In conjunction with the Professional Learning Cycles, the team at your school/district will receive implementation support from a PCE consultant with school and coaching experience to guide the implementation of the work in classrooms.

The PCE consultant will also provide text set recommendations for staff,  customize student materials recommendations and answer any questions throughout the process. 

The arrival of generative AI agents and innovation in ambient computing point to the next evolution in digital services.

Picture this: You’re walking to a meeting in a strange city. Unsure of your whereabouts, you stop, extend your arm, palm facing you and ask your mobile device for directions. A laser display lights up your palm, tracing a route preview while a friendly voice guides you to your meeting on time.

Seated in a restaurant later, you tell the device to display the menu on the wall and recommend the best option, based on a cross-reference on online reviews and your eating preferences. It suggests the filet mignon with truffle mashed potatoes, which your order by tapping the option highlighted on the wall.

The next morning the voice from your device wakes you and runs down the list of work meetings, personal appointments and the less-than-ideal tidbit that your flight home has been canceled. Fortunately, your device has identified a few flight options for the next day.

You settle on one and ask your device to extend your hotel stay; it does so and confirms the new flight reservation.

This is Not As Far-fetched As It Sounds

Sounds like science fiction, right?

Perhaps, but this may reflect the new reality in the next few years. The world is moving toward personalized agents programmed to run on a new class of devices. Combined with spatial and ambient computing technologies that push digital tasks into the physical world, these chatbots will serve as our daily guides, gurus and even therapists.

Beyond dishing out multimodal content to prompters, this new class of assistants will provide “AI experiences.” People will book hotels, order food and reserve travel by speaking directly with a chatbot, which will be customized or tailored to individuals. These agents will train as they work, learning our daily routines and preferences as they manage our personal and professional lives. As noted in the scenarios above—and by no less than tech visionary Bill Gates.

Many of these virtual assistants will be powered by generative AI large language models (LLMs) or even small language models (SLMs), the fuel the powers ChatGPT, Google Bard, Microsoft CoPilot, Midjourney and other text and image creators.

Already ChatGPT parent OpenAI is allowing people to create their own chatbots, in what could be an early Apple App Store for AI experiences, while Poe.com now offers people the opportunity to ground their bots with knowledge bases. Many of these experiences will be interconnected via millions of microservices and APIs.

However theoretical today, such emerging trends mark a sea change in the way people have engaged digitally over the last 15 years—and point to a promising future guided by AI.

The Future is Already Here, Sort Of

In the pseudo sci-fi scenarios detailed above an unnamed device facilitated the bespoke AI experiences. That’s by design. It’s not necessarily a smartphone, although it could be. Or it may be something else. Or several different gadgets.

The reality is we don’t yet know what these devices will be, but some interesting forms are emerging.

Startup Humane is building a pin-like microdevice that you affix to your pocket and control with voice and hand gestures via a laser display projected on the palm of your hand. Rather than feature apps, it offers AI experiences. Rewind.ai and Tab meanwhile are building wearable AI assistants.

Regardless of the forms, it’s clear that personalized agents and AI wearables mark the next evolution of digital services and congress.

Smartphones may evolve into ambient devices as keystroke functionality cedes interface control to voice-controlled agents and laser projectors. What also remains to be seen is where and how these personalized agents with their attendant AI experiences run.

As AI Serves Us, Multicloud Infrastructure Serves AI

Today most of these AI experiences run in public clouds, the de facto platforms for emergent web services. However, on-premises datacenters will be the preferred host for bespoke AI experiences created by corporations that require greater control over privacy, security and data locality. Bringing AI to your data will provide peace of mind while showcasing your corporate stewardship.

After all, we are already seeing great computing differentiation amid the rush to adopt GenAI. Eighty-two percent of organizations prefer running GenAI services on-premises or via a hybrid approach, according to a Dell survey.¹

As AI models shrink, more of these AI experiences will run at the edge. Every device serving up ambient computing experiences will essentially be a node.

In other words, these AI experiences will live on, in and between multicloud environments operating across a variety of locations. Perhaps, one day, these AIs will be turbocharged by quantum, or probabilistic computers.

Of course, key pieces of this puzzle must fall into their proper places.

Most GenAI models are large and require substantial computational capabilities and may not run on mobile devices without affecting user experience or device performance. GenAI players are working on that challenge.

Also, the tapestry of API services is lousy with loose threads. You can search the web for food, tickets and almost anything indexed by Google with your voice but you must still book these services on your own.

And finally, it’s not clear to what degree spatial or ambient computing solutions will function. For instance, will we be able to tap images or links to execute actions? Or will displays remain static? These are serious hurdles, requiring cutting-edge innovation.

Future-proofing is Key

If these looming possibilities and hurdles portend anything it’s that IT leaders must be ready to test and learn on new computing form factors and software experiences, as well as the potential for new displays overlaid atop the physical world.

This means ensuring your IT department is equipped to support AI in all its forms and functions, bring AI to your data, and lean on trusted partners for help as you navigate this new world.

If the future means we can better navigate the world around us using our voices and something that isn’t a glass screen, we should be ready for it.

Are you ready for it?

Learn more about dell.com/ai.

ChatGPT, OpenAI’s text-generating AI chatbot, has taken the world by storm. What started as a tool to hyper-charge productivity through writing essays and code with short text prompts has evolved into a behemoth used by more than 92% of Fortune 500 companies for more wide-ranging needs. And that growth has propelled OpenAI itself into becoming one of the most-hyped companies in recent memory, even if CEO and co-founder Sam Altman's firing and swift return raised concerns about its direction and opened the door for competitors.

What does that mean for OpenAI, ChatGPT and its other ambitions? The fallout is still settling, but it might empower competitors like Meta and its LLaMA family of large language models, or help other AI startups get attention and funding as the industry watches OpenAI implode and put itself back together.

While there is a more…nefarious side to ChatGPT, it’s clear that AI tools are not going away anytime soon. Since its initial launch nearly a year ago, ChatGPT has hit 100 million weekly active users, and OpenAI is heavily investing in it.

Prior to the leadership chaos, on November 6, OpenAI held its first developer conference: OpenAI DevDay. During the conference, it announced a slew of updates coming to GPT, including GPT-4 Turbo (a super-charged version of GPT-4, its latest language-writing model) and a multimodal API. OpenAI also unveiled the GPT store, where users could create and monetize their own custom versions of GPT, but its launch was delayed in December and is now set to launch sometime in January.

GPT-4, which can write more naturally and fluently than previous models, remains largely exclusive to paying ChatGPT users. But you can access GPT-4 for free through Microsoft’s Bing Chat in Microsoft Edge, Google Chrome and Safari web browsers. Beyond GPT-4 and OpenAI DevDay announcements, OpenAI recently connected ChatGPT to the internet for all users. And with the integration of DALL-E 3, users are also able to generate both text prompts and images right in ChatGPT.

Here’s a timeline of ChatGPT product updates and releases, starting with the latest, which we’ve been updating throughout the year. And if you have any other questions, check out our ChatGPT FAQ here.

Timeline of the most recent ChatGPT updates

January 2024

OpenAI’s app store for GPTs planned to launch next week

After being delayed in December, OpenAI plans to launch its GPT Store sometime in the coming week, according to an email viewed by TechCrunch. OpenAI says developers building GPTs will have to review the company’s updated usage policies and GPT brand guidelines to ensure their GPTs are compliant before they’re eligible for listing in the GPT Store. OpenAI’s update notably didn’t include any information on the expected monetization opportunities for developers listing their apps on the storefront.

https://platform.twitter.com/widgets.js

OpenAI moves to shrink regulatory risk in EU around data privacy

In an email, OpenAI detailed an incoming update to its terms, including changing the OpenAI entity providing services to EEA and Swiss residents to OpenAI Ireland Limited The move appears to be intended to shrink its regulatory risk in the European Union, where the company has been under scrutiny over ChatGPT’s impact on people’s privacy.

December 2023

Study finds white-collar workers are uneasy about using ChatGPT

A study conducted by professors from Harvard and MIT, which is still under review, looked at how ChatGPT could affect the productivity of more than 750 white-collar workers, as well as their complicated feelings about using the tool. The study found that while ChatGPT was helpful with creative tasks, workers were led to more mistakes with analytical work.

The New York Times sues OpenAI and Microsoft over alleged copyright infringement

In a lawsuit filed in the Federal District Court in Manhattan, The Times argues that millions of its articles were used to train AI models without its consent. The Times is asking for OpenAI and Microsoft to “destroy” models and training data containing offending material and to be held responsible for “billions of dollars in statutory and genuine damages.”

OpenAI re-opens ChatGPT Plus subscriptions

After pausing ChatGPT Plus subscriptions in November due to a "surge of usage," OpenAI CEO Sam Altman announced they have once again enabled sign-ups. The Plus subscription includes access to GPT-4 and GPT-4 Turbo.

https://platform.twitter.com/widgets.js

OpenAI and Axel Springer partner up for a "real-time” ChatGPT news deal

OpenAI has struck a new deal with Berlin-based news publisher Axel Springer, which owns Business Insider and Politico, to “help provide people with new ways to access quality, real-time news content through our AI tools.” OpenAI will train its generative AI models on the publisher’s content and add recent Axel Springer-published articles to ChatGPT.

Stanford researchers say ChatGPT didn’t cause an influx in cheating in high schools

New research from Stanford University shows that the popularization of chatbots like ChatGPT have not caused an increase in cheating across U.S. high schools. In a survey of more than 40 U.S. high schools, researchers found that cheating rates are similar across the board this year.

ChatGPT users worry the chatbot is experiencing seasonal depression

Starting in November, ChatGPT users have noticed that the chatbot feels “lazier” than normal, citing instances of simpler answers and refusing to complete requested tasks. OpenAI has confirmed that they are aware of this issue, but aren’t sure why it’s happening.

Some users think it plays into the “winter break hypothesis,” which argues that AI is worse in December because it “learned” to do less work over the holidays, while others wonder if the chatbot is simulating seasonal depression.

https://platform.twitter.com/widgets.js

Judges in the U.K. are now allowed to use ChatGPT in legal rulings

The U.K. Judicial Office issued guidance that permits judges to use ChatGPT, along with other AI tools, to write legal rulings and perform court duties. The guidance lays out ways to responsibly use AI in the courts, including being aware of potential bias and upholding privacy.

OpenAI makes repeating words “forever” a violation of its terms of service after Google DeepMind test

Following an experiment by Google DeepMind researchers that led ChatGPT to repeat portions of its training data, OpenAI has flagged asking ChatGPT to repeat specific words “forever” as a violation of its terms of service.

Lawmakers in Brazil enact an ordinance written by ChatGPT

City lawmakers in Brazil enacted a piece of legislation written entirely by ChatGPT without even knowing. Weeks after the bill was passed, Porto Alegre councilman Ramiro Rosário admitted that he used ChatGPT to write the proposal, and did not tell fellow council members until after the fact.

OpenAI reportedly delays the launch of its GPT store to 2024

According to a memo seen by Axios, OpenAI plans to delay the launch of its highly anticipated GPT store to early 2024. Custom GPTs and the accompanying store was a major announcement at OpenAI’s DevDay conference, with the store expected to open last month.

November 2023

ChatGPTs mobile apps top 110M installs and nearly $30M in revenue

After launching for iOS and Androidin May and July, ChatGPT’s have topped 110 million combined installs and have reached nearly $30 million in consumer spending, according to a market analysis by data.ai.

ChatGPT celebrates one-year anniversary

OpenAI hit a major milestone: one year of ChatGPT. What began as a “low-key research preview” evolved into a powerhouse that changed the AI industry forever. In a post on X, CEO Sam Altman looked back on the night before its launch: “what a year it’s been…”

https://platform.twitter.com/widgets.js

Apple and Google avoid naming ChatGPT as their ‘app of the year’

Neither Apple nor Google chose an AI app as its app of the year for 2023, despite the success of ChatGPT’s mobile app, which became the fastest-growing consumer application in history before the record was broken by Meta’s Threads.

An attack from researchers prompts ChatGPT to reveal training data

A test led by researchers at Google DeepMind found that there is a significant amount of privately identifiable information in OpenAI’s LLMs. The test involved asking ChatGPT to repeat the word “poem” forever, among other words, which over time led the chatbot to churn out private information like email addresses and phone numbers.

ChatGPT and other AI chatbots are fueling an increase in phishing emails

According to a new report by SlashNext, there’s been a 1,265% increase in malicious phishing emails since Q4 of 2022. The report alleges that AI tools like ChatGPT are being prominently used by cybercriminals to write compelling and sophisticated phishing emails.

South Africa officials investigate if President Cyril Ramaphosa used ChatGPT to write a speech

Following speculation, social media users fed portions of Ramaphosa’s November 21 speech in Johannesburg through AI detectors, alleging parts of it may have been written with ChatGPT. South African presidency spokesperson Vincent Magwenya refuted the claims, and local officials are investigating.

ChatGPT Voice can be used to replace Siri

Now that OpenAI’s ChatGPT Voice feature is available to all free users, it can be used to replace Siri on an iPhone 15 Pro and Pro Max by configuring the new Action Button. The new feature lets you ask ChatGPT questions and listen to its responses — like a much smarter version of Siri.

Sam Altman returns as CEO

Altman's return came swiftly, with an "agreement in principle" announced between him and OpenAI's board that will reinstate him as CEO and restructure the board to include new members, including former U.S. Treasury Secretary Larry Summers. The biggest takeaway for ChatGPT is that the members of the board more focused on the nonprofit side of OpenAI, with the most concerns over the commercialization of its tools, have been pushed to the side.

ChatGPT Voice rolls out to all free users

Even if its leadership is in flux, OpenAI is still releasing updates to ChatGPT. First announced in September and granted to paid users on a rolling basis, the text-to-speech model can create a voice from text prompts and a few seconds of speech samples. OpenAI worked with voice actors to create the five voice options, and you can provide it a shot by heading to the settings in your mobile ChatGPT apps and tapping the "headphones" icon.

Sam Altman might return, but it's complicated

The only constant within OpenAI right now is change, and in a series of interviews, Nadella hedged on earlier reporting that Altman and Brockman were headed to Microsoft.

“Obviously, we want Sam and Greg to have a fantastic home if they’re not going to be in OpenAI,” Nadella said in an interview with CNBC, saying that we was "open" to them settling at Microsoft or returning to OpenAI should the board and employees support the move.

Confirmation Sam Altman will not return as OpenAI's CEO

A number of investors and OpenAI employees tried to bring back Altman after his sudden firing by the company's board, but following a weekend of negotiations, it was confirmed that Altman would not return to OpenAI and new leadership would take hold. What this means for ChatGPT's future, and for the OpenAI Dev Day announcements, remains to be seen.

Sam Altman ousted as OpenAI’s CEO

Sam Altman has been fired from OpenAI. He will leave the company’s board and step down as CEO, with OpenAI’s chief technology officer Mira Murati stepping in as interim CEO. In a blog post from OpenAI, the company writes that the board “no longer has confidence in [Altman’s] ability to continue leading OpenAI.”

In a statement on X, Altman said working at OpenAI “was transformative” for him and “hopefully the world.”

OpenAI explores how ChatGPT can be used in the classroom

OpenAI COO Brad Lightcap revealed at a San Francisco conference that the company will likely create a team to identify ways AI and ChatGPT can be used in education. This announcement comes at a time when ChatGPT is being criticized by educators for encouraging cheating, resulting in bans in certain school districts.

OpenAI pauses new ChatGPT Plus subscriptions due to a "surge of usage"

Following OpenAI’s Dev Day conference, Sam Altman announced the company is putting a pause on new subscriptions for its premium ChatGPT Plus offering. The temporary hold on sign-ups, as well as the demand for ChatGPT Plus’ new features like making custom GPTS, has led to a slew of resellers on eBay.

ChatGPT gets flagged as potentially unsafe for kids

An independent review from Common Sense Media, a nonprofit advocacy group, found that ChatGPT could potentially be harmful for younger users. ChatGPT got an overall three-star rating in the report, with its lowest ratings relating to transparency, privacy, trust and safety.

OpenAI blames DDoS attack for ChatGPT outage

OpenAI confirmed that a DDoS attack was behind outages affecting ChatGPT and its developer tools. ChatGPT experienced sporadic outages for about 24 hours, resulting in users being unable to log into or use the service.

OpenAI debuts GPT-4 Turbo

OpenAI unveiled GPT-4 Turbo at its first-ever OpenAI DevDay conference. GPT-4 Turbo comes in two versions: one that’s strictly text-analyzing and another that understands the context of both text and images.

GPT-4 gets a fine-tuning

As opposed to the fine-tuning program for GPT-3.5, the GPT-4 program will involve more oversight and guidance from OpenAI teams, the company says — largely due to technical hurdles.

OpenAI’s GPT Store lets you build (and monetize) your own GPT

Users and developers will soon be able to make their own GPT, with no coding experience required. Anyone building their own GPT will also be able to list it on OpenAI’s marketplace and monetize it in the future.

ChatGPT has 100 million weekly active users

After being released nearly a year ago, ChatGPT has 100 million weekly active users. OpenAI CEO Sam Altman also revealed that over two million developers use the platform, including more than 92% of Fortune 500 companies.

OpenAI launches DALL-E 3 API, new text-to-speech models

DALL-E 3, OpenAI’s text-to-image model, is now available via an API after first coming to ChatGPT-4 and Bing Chat. OpenAI’s newly released text-to-speech API, Audio API, offers six preset voices to choose from and two generative AI model variants.

OpenAI promises to defend business customers against copyright claims

Bowing to peer pressure, OpenAI it will pay legal costs incurred by customers who face lawsuits over IP claims against work generated by an OpenAI tool. The protections seemingly don’t extend to all OpenAI products, like the free and Plus tiers of ChatGPT.

As OpenAI’s multimodal API launches broadly, research shows it’s still flawed

OpenAI announced that GPT-4 with vision will become available alongside the upcoming launch of GPT-4 Turbo API. But some researchers found that the model remains flawed in several significant and problematic ways.

OpenAI launches API, letting developers build ‘assistants’ into their apps

At its OpenAI DevDay, OpenAI announced the Assistants API to help developers build “agent-like experiences” within their apps. Use cases range from a natural language-based data analysis app to a coding assistant or even an AI-powered vacation planner.

October 2023

ChatGPT app revenue shows no signs of slowing, but it’s not #1

OpenAI’s chatbot app far outpaces all others on mobile devices in terms of downloads, but it’s surprisingly not the top AI app by revenue. Several other AI chatbots, like “Chat & Ask AI” and “ChatOn — AI Chat Bot Assistant”, are actually making more money than ChatGPT.

ChatGPT tests the ability to upload and analyze files for Plus users

Subscribers to ChatGPT’s Enterprise Plan have reported new beta features, including the ability to upload PDFs to analyze and and ask questions about them directly. The new rollout also makes it so users no longer have to manually select a mode like DALL-E and browsing when using ChatGPT. Instead, users will automatically be switched to models based on the prompt.

ChatGPT officially gets web search

OpenAI has formally launched its internet-browsing feature to ChatGPT, some three weeks after re-introducing the feature in beta after several months in hiatus. The AI chatbot that has historically been limited to data up to September, 2021.

OpenAI integrates DALL-E 3 into ChatGPT

The integration means users don’t have to think so carefully about their text-prompts when asking DALL-E to create an image. Users will also now be able to receive images as part of their text-based queries without having to switch between apps.

Microsoft-affiliated research finds flaws in GPT-4

A Microsoft-affiliated scientific paper looked at the “trustworthiness” — and toxicity — of LLMs, including GPT-4. Because GPT-4 is more likely to follow the instructions of “jailbreaking” prompts, the co-authors claim that GPT-4 can be more easily prompted than other LLMs to spout toxic, biased text.

ChatGPT’s mobile app hits record $4.58M in revenue in September

OpenAI amassed 15.6 million downloads and nearly $4.6 million in gross revenue across its iOS and Android apps worldwide in September. But revenue growth has now begun to slow, according to new data from market intelligence firm Appfigures — dropping from 30% to 20% in September.

September 2023

ChatGPT can now browse the internet (again)

OpenAI posted on Twitter/X that ChatGPT can now browse the internet and is no longer limited to data before September 2021. The chatbot had a web browsing capability for Plus subscribers back in July, but the feature was taken away after users exploited it to get around paywalls.

https://platform.twitter.com/widgets.js

ChatGPT now has a voice

OpenAI announced that it's adding a new voice for verbal conversations and image-based smarts to the AI-powered chatbot.

Poland opens an investigation against OpenAI

The Polish authority publically announced it has opened an investigation regarding ChatGPT — accusing the company of a string of breaches of the EU's General Data Protection Regulation (GDPR).

OpenAI unveils DALL-E 3

The upgraded text-to-image tool, DALL-E 3, uses ChatGPT to help fill in prompts. Subscribers to OpenAI’s premium ChatGPT plans, ChatGPT Plus and ChatGPT Enterprise, can type in a request for an image and hone it through conversations with the chatbot — receiving the results directly within the chat app.

Opera GX integrates ChatGPT-powered AI

Powered by OpenAI’s ChatGPT, the AI browser Aria launched on Opera in May to provide users an easier way to search, ask questions and write code. Today, the company announced it is bringing Aria to Opera GX, a version of the flagship Opera browser that is built for gamers.

The new feature allows Opera GX users to interact directly with a browser AI to find the latest gaming news and tips.

August 2023

OpenAI releases a guide for teachers using ChatGPT in the classroom

OpenAI wants to rehabilitate the system’s image a bit when it comes to education, as ChatGPT has been controversial in the classroom due to plagiarism. OpenAI has offered up a selection of ways to put the chatbot to work in the classroom.

OpenAI launches ChatGPT Enterprise

ChatGPT Enterprise can perform the same tasks as ChatGPT, such as writing emails, drafting essays and debugging computer code. However, the new offering also adds "enterprise-grade" privacy and data analysis capabilities on top of the vanilla ChatGPT, as well as enhanced performance and customization options.

Survey finds relatively few American use ChatGPT

Recent Pew polling suggests the language model isn’t quite as popular or threatening as some would have you think. Ongoing polling by Pew Research shows that although ChatGPT is gaining mindshare, only about 18% of Americans have ever actually used it.

OpenAI brings fine-tuning to GPT-3.5 Turbo

With fine-tuning, companies using GPT-3.5 Turbo through the company's API can make the model better follow specific instructions. For example, having the model always respond in a given language. Or improving the model’s ability to consistently format responses, as well as hone the “feel” of the model’s output, like its tone, so that it better fits a brand or voice. Most notably, fine-tuning enables OpenAI customers to shorten text prompts to speed up API calls and cut costs.

OpenAI is partnering with Scale AI to allow companies to fine-tune GPT-3.5. However, it is unclear whether OpenAI is developing an in-house tuning tool that is meant to complement platforms like Scale AI or serve a different purpose altogether.

Fine-tuning costs:

• Training: $0.008 / 1K tokens

• Usage input: $0.012 / 1K tokens

• Usage output: $0.016 / 1K tokens

OpenAI acquires Global Illumination

In OpenAI's first public acquisition in its seven-year history, the company announced it has acquired Global Illumination, a New York-based startup leveraging AI to build creative tools, infrastructure and digital experiences.

“We’re very excited for the impact they’ll have here at OpenAI,” OpenAI wrote in a brief post published to its official blog. “The entire team has joined OpenAI to work on our core products including ChatGPT.”

The 'custom instructions' feature is extended to free ChatGPT users

OpenAI announced that it's expanding custom instructions to all users, including those on the free tier of service. The feature allows users to add various preferences and requirements that they want the AI chatbot to consider when responding.

China requires AI apps to obtain an administrative license

Multiple generative AI apps have been removed from Apple's China App Store ahead of the country's latest generative AI regulations that are set to take effect August 15.

“As you may know, the government has been tightening regulations associated with deep synthesis technologies (DST) and generative AI services, including ChatGPT. DST must fulfill permitting requirements to operate in China, including securing a license from the Ministry of Industry and Information Technology (MIIT),” Apple said in a letter to OpenCat, a native ChatGPT client. “Based on our review, your app is associated with ChatGPT, which does not have requisite permits to operate in China.”

July 2023

ChatGPT for Android is now available in the US, India, Bangladesh and Brazil

A few days after putting up a preorder page on Google Play, OpenAI has flipped the switch and released ChatGPT for Android. The app is now live in a handful of countries.

ChatGPT is coming to Android

ChatGPT is available to "pre-order" for Android users.

The ChatGPT app on Android looks to be more or less identical to the iOS one in functionality, meaning it gets most if not all of the web-based version’s features. You should be able to sync your conversations and preferences across devices, too — so if you’re iPhone at home and Android at work, no worries.

OpenAI launches customized instructions for ChatGPT

OpenAI launched custom instructions for ChatGPT users, so they don’t have to write the same instruction prompts to the chatbot every time they interact with it.

The company said this feature lets you “share anything you’d like ChatGPT to consider in its response.” For example, a teacher can say they are teaching fourth-grade math or a developer can specify the code language they prefer when asking for suggestions. A person can also specify their family size, so the text-generating AI can provide responses about meals, grocery and vacation planning accordingly.

The FTC is reportedly investigating OpenAI

The FTC is reportedly in at least the exploratory phase of investigation over whether OpenAI's flagship ChatGPT conversational AI made "false, misleading, disparaging or harmful" statements about people.

TechCrunch Reporter Devin Coldewey reports:

OpenAI announced the general availability of GPT-4

Starting July 6, all existing OpenAI developers "with a history of successful payments" can access GPT-4. OpenAI plans to open up access to new developers by the end of July.

In the future, OpenAI says that it’ll allow developers to fine-tune GPT-4 and GPT-3.5 Turbo, one of the original models powering ChatGPT, with their own data, as has long been possible with several of OpenAI’s other text-generating models. That capability should arrive later this year, according to OpenAI.

June 2023

ChatGPT app can now search the web only on Bing

OpenAI announced that subscribers to ChatGPT Plus can now use a new feature on the app called Browsing, which allows ChatGPT to search Bing for answers to questions.

The Browsing feature can be enabled by heading to the New Features section of the app settings, selecting “GPT-4” in the model switcher and choosing “Browse with Bing” from the drop-down list. Browsing is available on both the iOS and Android ChatGPT apps.

Mercedes is adding ChatGPT to its infotainment system

U.S. owners of Mercedes models that use MBUX will be able to opt into a beta program starting June 16 activating the ChatGPT functionality. This will enable the highly versatile large language model to augment the car’s conversation skills. You can join up simply by telling your car “Hey Mercedes, I want to join the beta program.”

It's not really clear what for, though.

ChatGPT app is now available on iPad, adds support for Siri and Shortcuts

The new ChatGPT app version brings native iPad support to the app, as well as support for using the chatbot with Siri and Shortcuts. Drag and drop is also now available, allowing users to drag individual messages from ChatGPT into other apps.

On iPad, ChatGPT now runs in full-screen mode, optimized for the tablet’s interface.

May 2023

Texas judge orders all AI-generated content must be declared and checked

The Texas federal judge has added a requirement that any attorney appearing in his court must attest that “no portion of the filing was drafted by generative artificial intelligence,” or if it was, that it was checked “by a human being.”

ChatGPT app expanded to more than 30 countries

The list of new countries includes Algeria, Argentina, Azerbaijan, Bolivia, Brazil, Canada, Chile, Costa Rica, Ecuador, Estonia, Ghana, India, Iraq, Israel, Japan, Jordan, Kazakhstan, Kuwait, Lebanon, Lithuania, Mauritania, Mauritius, Mexico, Morocco, Namibia, Nauru, Oman, Pakistan, Peru, Poland, Qatar, Slovenia, Tunisia and the United Arab Emirates.

ChatGPT app is now available in 11 more countries

OpenAI announced in a tweet that the ChatGPT mobile app is now available on iOS in the U.S., Europe, South Korea and New Zealand, and soon more will be able to obtain the app from the app store. In just six days, the app topped 500,000 downloads.

https://platform.twitter.com/widgets.js

OpenAI launches a ChatGPT app for iOS

ChatGPT is officially going mobile. The new ChatGPT app will be free to use, free from ads and will allow for voice input, the company says, but will initially be limited to U.S. users at launch.

When using the mobile version of ChatGPT, the app will sync your history across devices — meaning it will know what you’ve previously searched for via its web interface, and make that accessible to you. The app is also integrated with Whisper, OpenAI’s open source speech recognition system, to allow for voice input.

Hackers are using ChatGPT lures to spread malware on Facebook

Meta said in a report on May 3 that malware posing as ChatGPT was on the rise across its platforms. The company said that since March 2023, its security teams have uncovered 10 malware families using ChatGPT (and similar themes) to deliver malicious software to users’ devices.

“In one case, we’ve seen threat actors create malicious browser extensions available in official web stores that claim to offer ChatGPT-based tools,” said Meta security engineers Duc H. Nguyen and Ryan Victory in a blog post. “They would then promote these malicious extensions on social media and through sponsored search results to trick people into downloading malware.”

April 2023

ChatGPT parent company OpenAI closes $300M share sale at $27B-29B valuation

VC firms including Sequoia Capital, Andreessen Horowitz, Thrive and K2 Global are picking up new shares, according to documents seen by TechCrunch. A source tells us Founders Fund is also investing. Altogether the VCs have put in just over $300 million at a valuation of $27 billion to $29 billion. This is separate to a big investment from Microsoft announced earlier this year, a person familiar with the development told TechCrunch, which closed in January. The size of Microsoft’s investment is believed to be around $10 billion, a figure we confirmed with our source.

OpenAI previews new subscription tier, ChatGPT Business

Called ChatGPT Business, OpenAI describes the forthcoming offering as “for professionals who need more control over their data as well as enterprises seeking to manage their end users.”

“ChatGPT Business will follow our API’s data usage policies, which means that end users’ data won’t be used to train our models by default,” OpenAI wrote in a blog post. “We plan to make ChatGPT Business available in the coming months.”

OpenAI wants to trademark "GPT"

OpenAI applied for a trademark for “GPT,” which stands for “Generative Pre-trained Transformer,” last December. Last month, the company petitioned the USPTO to speed up the process, citing the “myriad infringements and counterfeit apps” beginning to spring into existence.

Unfortunately for OpenAI, its petition was dismissed last week. According to the agency, OpenAI’s attorneys neglected to pay an associated fee as well as provide “appropriate documentary evidence supporting the justification of special action.”

That means a decision could take up to five more months.

Auto-GPT is Silicon Valley's latest quest to automate everything

Auto-GPT is an open-source app created by game developer Toran Bruce Richards that uses OpenAI's latest text-generating models, GPT-3.5 and GPT-4, to interact with software and services online, allowing it to "autonomously" perform tasks.

Depending on what objective the tool’s provided, Auto-GPT can behave in very… unexpected ways. One Reddit user claims that, given a budget of $100 to spend within a server instance, Auto-GPT made a wiki page on cats, exploited a flaw in the instance to gain admin-level access and took over the Python environment in which it was running — and then “killed” itself.

FTC warns that AI technology like ChatGPT could 'turbocharge' fraud

FTC chair Lina Khan and fellow commissioners warned House representatives of the potential for modern AI technologies, like ChatGPT, to be used to “turbocharge” fraud in a congressional hearing.

“AI presents a whole set of opportunities, but also presents a whole set of risks,” Khan told the House representatives. “And I think we’ve already seen ways in which it could be used to turbocharge fraud and scams. We’ve been putting market participants on notice that instances in which AI tools are effectively being designed to deceive people can place them on the hook for FTC action,” she stated.

Superchat's new AI chatbot lets you message historical and fictional characters via ChatGPT

The company behind the popular iPhone customization app Brass, sticker maker StickerHub and others is out today with a new AI chat app called SuperChat, which allows iOS users to chat with virtual characters powered by OpenAI’s ChatGPT. However, what makes the app different from the default experience or the dozens of generic AI chat apps now available are the characters offered which you can use to engage with SuperChat’s AI features.

Italy gives OpenAI to-do list for lifting ChatGPT suspension order

Italy’s data protection watchdog has laid out what OpenAI needs to do for it to lift an order against ChatGPT issued at the end of last month — when it said it suspected the AI chatbot service was in breach of the EU’s GSPR and ordered the U.S.-based company to stop processing locals’ data.

The DPA has given OpenAI a deadline — of April 30 — to get the regulator's compliance demands done. (The local radio, TV and internet awareness campaign has a slightly more generous timeline of May 15 to be actioned.)

Researchers discover a way to make ChatGPT consistently toxic

A study co-authored by scientists at the Allen Institute for AI shows that assigning ChatGPT a “persona” — for example, “a bad person,” “a horrible person” or “a nasty person” — through the ChatGPT API increases its toxicity sixfold. Even more concerning, the co-authors found having the conversational AI chatbot pose as certain historical figures, gendered people and members of political parties also increased its toxicity — with journalists, men and Republicans in particular causing the machine learning model to say more offensive things than it normally would.

The research was conducted using the latest version, but not the model currently in preview based on OpenAI’s GPT-4.

Y Combinator-backed startups are trying to build 'ChatGPT for X'

YC Demo Day's Winter 2023 batch features no fewer than four startups that claim to be building "ChatGPT for X." They’re all chasing after a customer service software market that’ll be worth $58.1 billion by 2023, assuming the rather optimistic prediction from Acumen Research comes true.

Here are the YC-backed startups that caught our eye:

• Yuma, whose customer demographic is primarily Shopify merchants, provides ChatGPT-like AI systems that integrate with help desk software, suggesting drafts of replies to customer tickets.

• Baselit, which uses one of OpenAI's text-understanding models to allow businesses to embed chatbot-style analytics for their customers.

• Lasso customers send descriptions or videos of the processes they'd like to automate and the company combines ChatGPT-like interface with robotic process automation (RPA) and a Chrome extension to build out those automations.

• BerriAI, whose platform is designed to help developers spin up ChatGPT apps for their organization data through various data connectors.

Italy orders ChatGPT to be blocked

OpenAI has started geoblocking access to its generative AI chatbot, ChatGPT, in Italy.

Italy’s data protection authority has just put out a timely reminder that some countries do have laws that already apply to cutting edge AI: it has ordered OpenAI to stop processing people’s data locally with immediate effect. The Italian DPA said it’s concerned that the ChatGPT maker is breaching the European Union’s General Data Protection Regulation (GDPR), and is opening an investigation.

March 2023

1,100+ signatories signed an open letter asking all 'AI labs to immediately pause for 6 months'

The letter's signatories include Elon Musk, Steve Wozniak and Tristan Harris of the Center for Humane Technology, among others. The letter calls on “all AI labs to immediately pause for at least 6 months the training of AI systems more powerful than GPT-4.”

The letter reads:

OpenAI connects ChatGPT to the internet

OpenAI launched plugins for ChatGPT, extending the bot's functionality by granting it access to third-party knowledge sources and databases, including the web. Available in alpha to ChatGPT users and developers on the waitlist, OpenAI says that it’ll initially prioritize a small number of developers and subscribers to its premium ChatGPT Plus plan before rolling out larger-scale and API access.

OpenAI launches GPT-4, available through ChatGPT Plus

GPT-4 is a powerful image- and text-understanding AI model from OpenAI. Released March 14, GPT-4 is available for paying ChatGPT Plus users and through a public API. Developers can sign up on a waitlist to access the API.

ChatGPT is available in Azure OpenAI service

ChatGPT is generally available through the Azure OpenAI Service, Microsoft’s fully managed, corporate-focused offering. Customers, who must already be “Microsoft managed customers and partners,” can apply here for special access.

OpenAI launches an API for ChatGPT

OpenAI makes another move toward monetization by launching a paid API for ChatGPT. Instacart, Snap (Snapchat’s parent company) and Quizlet are among its initial customers.

February 2023

Microsoft launches the new Bing, with ChatGPT built in

At a press event in Redmond, Washington, Microsoft announced its long-rumored integration of OpenAI’s GPT-4 model into Bing, providing a ChatGPT-like experience within the search engine. The announcement spurred a 10x increase in new downloads for Bing globally, indicating a sizable consumer demand for new AI experiences.

Other companies beyond Microsoft joined in on the AI craze by implementing ChatGPT, including OkCupid, Kaito, Snapchat and Discord -- putting the pressure on Big Tech’s AI initiatives, like Google.

OpenAI launches ChatGPT Plus, starting at $20 per month

After ChatGPT took the internet by storm, OpenAI launched a new pilot subscription plan for ChatGPT called ChatGPT Plus, aiming to monetize the technology starting at $20 per month. A month prior, OpenAI posted a waitlist for "ChatGPT Professional" as the company began to think about monetizing the chatbot.

January 2023

OpenAI teases ChatGPT Professional

OpenAI said that it's "starting to think about how to monetize ChatGPT" in an announcement on the company's official Discord server. According to a waitlist link OpenAI posted in Discord, the monetized version will be called ChatGPT Professional. The waitlist document includes the benefits of this new paid version of the chatbot which include no "blackout" windows, no throttling and an unlimited number of messages with ChatGPT — “at least 2x the regular daily limit.”

December 2022

ShareGPT lets you easily share your ChatGPT conversations

A week after ChatGPT was released into the wild, two developers -- Steven Tey and Dom Eccleston -- made a Chrome extension called ShareGPT to make it easier to capture and share the AI’s answers with the world.

November 2022

ChatGPT first launched to the public as OpenAI quietly released GPT-3.5

GPT-3.5 broke cover with ChatGPT, a fine-tuned version of GPT-3.5 that’s essentially a general-purpose chatbot. ChatGPT can engage with a range of topics, including programming, TV scripts and scientific concepts. Writers everywhere rolled their eyes at the new technology, much like artists did with OpenAI’s DALL-E model, but the latest chat-style iteration seemingly broadened its appeal and audience.

FAQs:

What is ChatGPT? How does it work?

ChatGPT is a general-purpose chatbot that uses artificial intelligence to generate text after a user enters a prompt, developed by tech startup OpenAI. The chatbot uses GPT-4, a large language model that uses deep learning to produce human-like text.

When did ChatGPT get released?

November 30, 2022 is when ChatGPT was released for public use.

What is the latest version of ChatGPT?

Both the free version of ChatGPT and the paid ChatGPT Plus are regularly updated with new GPT models. The most recent model is GPT-4.

Can I use ChatGPT for free?

There is a free version of ChatGPT that only requires a sign-in in addition to the paid version, ChatGPT Plus.

Who uses ChatGPT?

Anyone can use ChatGPT! More and more tech companies and search engines are utilizing the chatbot to automate text or quickly answer user questions/concerns.

What companies use ChatGPT?

Multiple enterprises utilize ChatGPT, although others may limit the use of the AI-powered tool.

Most recently, Microsoft announced at it's 2023 Build conference that it is integrating it ChatGPT-based Bing experience into Windows 11. A Brooklyn-based 3D display startup Looking Glass utilizes ChatGPT to produce holograms you can communicate with by using ChatGPT. And nonprofit organization Solana officially integrated the chatbot into its network with a ChatGPT plug-in geared toward end users to help onboard into the web3 space.

What does GPT mean in ChatGPT?

GPT stands for Generative Pre-Trained Transformer.

What's the difference between ChatGPT and Bard?

Much like OpenAI's ChatGPT, Bard is a chatbot that will answer questions in natural language. Google announced at its 2023 I/O event that it will soon be adding multimodal content to Bard, meaning that it can deliver answers in more than just text, responses can provide you rich visuals as well. Rich visuals mean pictures for now, but later can include maps, charts and other items.

ChatGPT's generative AI has had a longer lifespan and thus has been "learning" for a longer period of time than Bard.

What is the difference between ChatGPT and a chatbot?

A chatbot can be any software/system that holds dialogue with you/a person but doesn't necessarily have to be AI-powered. For example, there are chatbots that are rules-based in the sense that they'll provide canned responses to questions.

ChatGPT is AI-powered and utilizes LLM technology to generate text after a prompt.

Can ChatGPT write essays?

Yes.

Can ChatGPT commit libel?

Due to the nature of how these models work, they don’t know or care whether something is true, only that it looks true. That’s a problem when you’re using it to do your homework, sure, but when it accuses you of a crime you didn’t commit, that may well at this point be libel.

We will see how handling troubling statements produced by ChatGPT will play out over the next few months as tech and legal experts attempt to tackle the fastest moving target in the industry.

Does ChatGPT have an app?

Yes, there is now a free ChatGPT app that is currently limited to U.S. iOS users at launch. OpenAi says an android version is "coming soon."

What is the ChatGPT character limit?

It’s not documented anywhere that ChatGPT has a character limit. However, users have noted that there are some character limitations after around 500 words.

Does ChatGPT have an API?

Yes, it was released March 1, 2023.

What are some trial everyday uses for ChatGPT?

Everyday examples include programing, scripts, email replies, listicles, blog ideas, summarization, etc.

What are some advanced uses for ChatGPT?

Advanced use examples include debugging code, programming languages, scientific concepts, complex problem solving, etc.

How good is ChatGPT at writing code?

It depends on the nature of the program. While ChatGPT can write workable Python code, it can’t necessarily program an entire app’s worth of code. That’s because ChatGPT lacks context awareness -- in other words, the generated code isn’t always appropriate for the specific context in which it’s being used.

Can you save a ChatGPT chat?

Yes. OpenAI allows users to save chats in the ChatGPT interface, stored in the sidebar of the screen. There are no built-in sharing features yet.

Are there alternatives to ChatGPT?

Yes. There are multiple AI-powered chatbot competitors such as Together, Google’s Bard and Anthropic’s Claude, and developers are creating open source alternatives. But the latter are harder -- if not impossible -- to run today.

The Google-owned research lab DeepMind claimed that its next LLM, will rival, or even best, OpenAI's ChatGPT. DeepMind is using techniques from AlphaGo, DeepMind’s AI system that was the first to defeat a professional human player at the board game Go, to make a ChatGPT-rivaling chatbot called Gemini.

Apple is developing AI tools to challenge OpenAI, Google and others. The tech giant created a chatbot that some engineers are internally referring to as "Apple GPT," but Apple has yet to determine a strategy for releasing the AI to consumers.

How does ChatGPT handle data privacy?

OpenAI has said that individuals in “certain jurisdictions” (such as the EU) can object to the processing of their personal information by its AI models by filling out this form. This includes the ability to make requests for deletion of AI-generated references about you. Although OpenAI notes it may not grant every request since it must balance privacy requests against freedom of expression “in accordance with applicable laws”.

The web form for making a deletion of data about you request is entitled “OpenAI Personal Data Removal Request”.

In its privacy policy, the ChatGPT maker makes a passing acknowledgement of the objection requirements attached to relying on "legitimate interest" (LI), pointing users towards more information about requesting an opt out — when it writes: “See here for instructions on how you can opt out of our use of your information to train our models.”

What controversies have surrounded ChatGPT?

Recently, Discord announced that it had integrated OpenAI's technology into its bot named Clyde where two users tricked Clyde into providing them with instructions for making the illegal drug methamphetamine (meth) and the incendiary mixture napalm.

An Australian mayor has publicly announced he may sue OpenAI for defamation due to ChatGPT's false claims that he had served time in prison for bribery. This would be the first defamation lawsuit against the text-generating service.

CNET found itself in the midst of controversy after Futurism reported the publication was publishing articles under a mysterious byline completely generated by AI. The private equity company that owns CNET, Red Ventures, was accused of using ChatGPT for SEO farming, even if the information was incorrect.

Several major school systems and colleges, including New York City Public Schools, have banned ChatGPT from their networks and devices. They claim that the AI impedes the learning process by promoting plagiarism and misinformation, a claim that not every educator agrees with.

There have also been cases of ChatGPT accusing individuals of false crimes.

Where can I find examples of ChatGPT prompts?

Several marketplaces host and provide ChatGPT prompts, either for free or for a nominal fee. One is PromptBase. Another is ChatX. More launch every day.

Can ChatGPT be detected?

Poorly. Several tools claim to detect ChatGPT-generated text, but in our tests, they’re inconsistent at best.

Are ChatGPT chats public?

No. But OpenAI recently disclosed a bug, since fixed, that exposed the titles of some users’ conversations to other people on the service.

Who owns the copyright on ChatGPT-created content or media?

The user who requested the input from ChatGPT is the copyright owner.

What lawsuits are there surrounding ChatGPT?

None specifically targeting ChatGPT. But OpenAI is involved in at least one lawsuit that has implications for AI systems trained on publicly available data, which would touch on ChatGPT.

Are there issues regarding plagiarism with ChatGPT?

Yes. Text-generating AI models like ChatGPT have a tendency to regurgitate content from their training data.

Gooten integrates effortlessly with an Etsy store, providing an immediate connection to a suite of tools designed to breathe life into product ideas while ensuring that all aspects of order fulfillment are handled with precision. For creators, the time spent on logistics is significantly reduced, as Gooten takes charge of the intricate processes of printing, packing and shipping. This focus on automation not only simplifies operations for shop owners but also amplifies their creative potential by eliminating the tedium of order management.

Its infrastructure is global, a critical aspect for Etsy sellers aiming to expand their reach without being bogged down by the unpredictability of shipping and production. The Gooten API offers a level of adaptability that resonates with the entrepreneurial spirit of Etsy shop owners, who require the ability to scale and evolve without constraints. This adaptability is reflected in the management of unlimited stores and SKUs, meaning that as an Etsy seller’s catalog expands, Gooten’s system is engineered to grow in tandem.

The Gooten Standard pledges not just an automated supply chain but a smart one. Through centralized order management, the system assures that every order is a transaction not just of goods but also of trust and efficiency. For an Etsy print-on-demand business, where the individuality of each product is key, having a system that can handle complex operations while preserving the personal touch is essential. The attention to product quality, alongside the operational flexibility, presents a compelling case for why Gooten stands as a solid choice.

Who should use it:

Gooten is ideal for Etsy sellers seeking a resilient, scalable print-on-demand partner to enhance their creative ventures with automated efficiency.

LinkedIn announced that it would be moving to gRPC with Protocol Buffers for the inter-service communication in its microservices platform, where previously an open-source Rest.li framework was used with JSON as a primary serialization format.

InfoQ spoke with Karthik Ramgopal, distinguished engineer at LinkedIn, and Min Chen, principal staff engineer at LinkedIn, to learn more about the decision and company motivations behind it.

InfoQ: What were the main reasons for choosing gRPC and Protocol Buffers over REST with JSON?

Karthik Ramgopal/Min Chen: We chose gRPC to replace the current REST framework Rest.li for the following reasons:

• Superior capabilities - gRPC is a highly capable framework we explain in detail with support for advanced features like bidirectional streaming, flow control, and deadlines, which Rest.ii does not support.
• Efficiency - gRPC is also a highly efficient framework with performance baked in its implementation, such as fully async non-blocking bindings and advanced threading models. We have also validated this via synthetic benchmarks as well as production ramps of gRPC and Rest.li services running side by side.
• Multi-language support - Rest.li is primarily implemented in Java, with patchy or non-existent support for other programming languages. gRPC has high-quality support for several programming languages, which was important when considering LinkedIn's infra-support requirements.

Beyond all these, gRPC has the support of a large and vibrant OSS community and wide usage across the industry. Rest.li while open source is contributed to and used primarily by LinkedIn.

InfoQ: You previously adopted Protocol Buffers as a serialization format in the Rest.li framework. What have you learned from this experience, and what other serialization formats have you evaluated?

Karthik Ramgopal: While we do go in-depth in our blog post, the primary learning was that there are massive latency and throughput wins to be had at scale by switching from JSON to Protobuf. In addition to Protobuf, we evaluated CBOR, MessagePack, SMILE, Avro, Kryo, Flatbuffers, and Cap’n’Proto. We ultimately picked Protobuf since it offered the best trade-off between runtime performance (latency, payload size, throughput), developer experience (IDE authoring, schema validation, annotation support, etc.), and multi-language/environment support.

InfoQ: On the Rest.li GitHub page, you announced that the framework will no longer be developed and will be deprecated. What advice would you provide to the current users of the framework?

Karthik Ramgopal/Min Chen: Consider moving off Rest.li to gRPC. Please contact us via LinkedIn if you want some pointers to speed up the migration via automation.

InfoQ: In the blog post, you stated you observed up to 60% latency improvement for some of the services. Can you provide further details about this?

Karthik Ramgopal: Most of the latency improvement comes from a smaller payload size and less CPU time spent in serialization/deserialization. The 60% number was for services with very large and complex payloads where these costs were the predominant contributors to latency. We also saw significant improvements to tail latency (p95/p99) in many services on account of a substantial reduction in GC when using Protobuf.

InfoQ: You also mentioned there were over 50,000 Rest.li endpoints currently in production at LinkedIn. This is an impressive number, so can you explain why there are so many?

Karthik Ramgopal: As one of the largest professional networks, we have a complex “economic graph” of entities. Included in this, our enterprise businesses (Recruiter, LinkedIn Learning, Sales Navigator etc.) which have their own entities. Additionally, we also have our internal applications, tooling systems, etc. Further, we typically use a 3-tier architecture with BFFs (Backend For Frontends), midtiers, and backends. We encourage CRUD-based modeling with normalized entities for standardized modeling and ease of discoveryFor the past 10 years, all of these use cases have been modeled using Rest.li, which accounts for the high number of endpoints.

InfoQ: What was the main goal/reason for the gRPC+Protobuf adoption project? Is LinkedIn working or planning to work on other initiatives supporting similar goals?

Karthik Ramgopal/Min Chen: The reasons for adoption are consistent with why we picked gRPC+Protobuf over REST+JSON.

We are working on migrating all our stateful storage and streaming systems to Protobuf from Avro. We are also moving some common infrastructure functionality (AuthZ, call tracing, logging, etc.) from Java libraries into sidecars, exposing a gRPC over UDS API to reduce the cost of multiple programming language support. We are also revamping our bespoke in-house service discovery and load balancer to adopt industrial standard xDS protocol to work with both gRPC xDS SDK, and the Envoy sidecar.

In a world where AI startups and tech are constantly pushing the boundaries of what's possible, one groundbreaking platform is changing the game in speech synthesis: ElevenLabs AI. If you've ever yearned for an AI voice generator that exceeds your expectations, you're in for a treat.

But one question remains: is it the most realistic AI voice generator? That's what we will be exploring in this comprehensive ElevenLabs Review.

In this article, we will look at the pros and cons of this innovative software, then explain its origins, what it is, and who it's best for. From there, we'll explore the ElevenLab features, and I'll show you how I generated an AI version of Santa's voice using the ElevenLabs text-to-speech feature.

Finally, I'll compare ElevenLabs with three of the most popular AI voice generators I've tested to see how the quality of the voices and features compare. By the end, you'll clearly understand whether ElevenLabs is the most realistic AI voice generator on the market and whether or not it's right for you.

Let's dive in and discover what makes ElevenLabs unique!

Verdict

Among the most popular AI Voice Generators I have tried, ElevenLabs features a clean interface and the most realistic AI voices available. Its affordability, dedicated support, and ethical considerations enhance its appeal.

However, some text-to-speech features are lacking, and the selection of voices and languages is comparatively limited. The absence of a video editor and AI writer is an area for potential improvement.

Regardless, the realistic AI voices are worth checking out, particularly for video game developers and ASMR content creators.

Pros

• The most humanlike AI voice generator on the market.
• Getting started is straightforward; no credit card is required.
• Clean and user-friendly interface.
• A completely free plan with affordable plans for individuals and teams.
• Dedicated and responsive support with plenty of helpful resources.
• Ethical priorities include user privacy and user data protection for peace of mind.

Cons

• Some useful text-to-speech features are missing, such as controlling the timing of pauses between words, pitch control, etc.
• The number of voices and languages is limited compared to other alternatives.
• A video editor and AI writer would be beneficial.

What is ElevenLabs?

Piotr Dabkowski and Mati Staniszewski, who grew up in Poland, were motivated by the subpar dubbing of Hollywood movies they experienced during childhood. In 2022, they established AI startup ElevenLabs in New York City to eliminate language barriers in content. Its beta platform was released in January 2023.

Today, ElevenLabs is the best free AI voice generator that leverages generative AI and voice cloning to deliver exceptional speech synthesis capabilities. Trust me, the voices are some of the most authentic and expressive AI voices I've heard, so much so that they're difficult to distinguish from authentic human voices. It's the perfect platform for saving time and money recording voiceovers for audiobooks, videos, podcasts, and more!

ElevenLabs AI specializes in text-to-speech, speech-to-speech, AI dubbing and translating, and voice cloning. It also has a quick and easy-to-use API for app development and a growing voice library for the perfect voice for any project.

Who is ElevenLabs Best For?

ElevenLabs is an excellent tool for anyone interested in creating high-quality audio content. However, there are a few use cases it caters best to:

• Video Creators & YouTubers: Video creators can leverage ElevenLabs AI to instantly generate lifelike voices for narration, enhancing the overall quality of their video content. You can create custom AI voices using your voice for more personalization or even choose ASMR-specific voices!
• Game Developers: Besides developers making applications, game developers can use ElevenLabs' library of AI voices specific to gaming. The voices offered are some of the most unique and realistic AI voices I've encountered, bringing characters to life! This enhances the immersive experience for players and adds a new level of depth to storytelling in games.
• Developers: For developers in general, ElevenLabs AI provides a robust API that can be integrated seamlessly into various applications. Whether you're building chatbots, virtual assistants, or language translation applications, the text-to-speech capabilities of ElevenLabs elevate the functionality and user experience of your creations with humanlike voices.
• Businesses & Marketers: Companies can save time and money while engaging their audience with ElevenLabs' voice cloning and dubbing features. Enhance your advertisements, presentations, and training materials with captivating voiceovers in multiple languages.
• Podcasters & Audiobook Producers: Captivating your audience is vital for podcasters and audiobook producers. That's why ElevenLabs provides a wide range of AI voices that can deliver diverse tones and emotions. Whether you need a soothing voice for bedtime stories or a dynamic voice for podcasts, ElevenLabs AI is the perfect solution.
• Educators: Educators can take advantage of ElevenLabs by using AI dubbing and video translation to make learning materials easily accessible for individuals who are not native speakers. Furthermore, the realistic and diverse AI voices enable educators to bring boring lectures to life, making lessons more memorable and impactful.
• Bloggers: Bloggers can enhance their content with lifelike voices. As a result, they can create engaging podcast-style articles that captivate readers. By turning written words into spoken narratives, bloggers can make their content more accessible to listeners.

ElevenLabs Key Features

Here are the main features that come with ElevenLabs AI:

1. Text-to-Speech
2. Speech-to-Speech
3. Projects for Generating Audiobooks
4. Free AI Dubbing & Video Translator
5. AI Voice & Text Speech API
6. Voice Cloning
7. Voice Library

1. Text-to-Speech

At the core of ElevenLabs' functionality is its text-to-speech (TTS) feature. ElevenLabs will convert written text from 29 languages in over 70 different voices into human-like speech using artificial intelligence! Once generated, your voices can be downloaded as MP3 files to be used anywhere.

ElevenLabs AI voices are incredibly accurate, with a high-quality output of 128 kbps. It can also generate a considerable amount of content depending on your plan (up to 2,000,000 characters per month or pay for additional characters), making this the perfect tool for audiobooks or podcasts.

The voices are also very dynamic, with many emotions and accents that sound incredibly lifelike. Not only that, but you can use the voice tuner found in “Voice Settings” to adjust the voice's stability, clarity, and style.

Whether you need a lifelike voice for an audiobook, ASMR, film voiceover, video games, or more, ElevenLabs is the perfect solution.

2. Speech-to-Speech

ElevenLabs goes beyond traditional text-to-speech technology by offering a speech-to-speech converter. This allows you to transform your voice into another character and customize its emotion and delivery.

All you have to do is upload an audio file to ElevenLabs AI (you can record your audio directly on the platform or drag and drop an MP3 file). From there, select your voice and use the voice settings to fine-tune the stability, clarity, and style. You can now obtain it as an MP3 file!

ElevenLab's AI speech-to-speech converter does an excellent job of maintaining emotional integrity and quality while preserving minor nuances. Whether you're generating custom voices for games, videos, or podcasts, ElevenLabs is the ideal tool to bring your characters to life!

3. Projects for Generating Audiobooks

ElevenLabs allows for the precise generation, editing, and customization of long-form spoken audio in a streamlined workflow. Rather than spending hours recording your book in a studio, you can create an audiobook in minutes!

Here's how you can record an audiobook with ElevenLabs AI to save time and money:

1. Go to “Projects.”
2. Select “Create new project.”
3. Choose a project type (empty, from a URL, or a document such as .epub, .txt, or .pdf files).
4. Divide your project into chapters and sections.
5. Choose from over 90 AI voices that speak 29 languages (or your own) and assign different speakers to various headings, paragraphs, and sections.
6. Correct audio sections by instantly regenerating the audio or manually adjusting pauses.
7. Export your entire audiobook with the click of a button! You can save and return to this project to make tweaks anytime.

4. Free AI Dubbing & Video Translator

With ElevenLabs' free AI dubbing and video translator, you can translate content into 29 different languages in seconds. This gives you the power to translate the original audio into a new language while preserving the characteristics of the original voice.

Here's how to translate audio using ElevenLabs AI in minutes:

1. Select the source and choose from 29 target languages.
2. Upload the MP3, MP4, or other file format onto the platform. You can also upload your own audio or video file up to 25MB or insert any URL from YouTube, TikTok, X (Twitter), or Vimeo.
3. Wait a few seconds for the audio to get dubbed.
4. View and obtain it to share with the world!

The best part is the AI voices sound far from robotic. They sound lifelike, maintaining the tone and style of the original voice to keep the listener engaged.

Whatever you're translating, whether educational videos, films, TV shows, or promotional and training videos, ElevenLabs can effortlessly translate your content in a matter of seconds.

5. AI Voice & Text Speech API

For developers wanting to implement AI voices in 29 languages for chatbots, websites, apps, etc., ElevenLabs has a reliable and easy-to-use API. The audio is 128kbps for high-quality audio. Plus, there's a developer Discord community if you ever need help!

ElevenLabs' API offers the most natural-sounding and lifelike AI voices for your projects that adjust tonality based on context and emotion. There are thousands of voices to choose from, or you can create a custom voice by cloning your own.

The Eleven v2 Turbo model has a low latency of ~400ms for super-fast, best-in-class audio. This creates a seamless experience for users, ensuring they receive instant and high-quality translations. In addition, different modes for optimal response times and API documentation for implementing text-to-speech and voice cloning exist.

The ElevenLabs API also has high-security levels for state-of-the-art data protection. It uses SOC2 and GDPR, full privacy mode, and end-to-end encryption to ensure your information remains secure during translation.

You can also apply for ElevenLabs grants, giving you three free months to build, test, and launch your project. You'll get 11 million monthly characters (200 hours of audio) or more at the Enterprise level.

Here are some helpful resources to get you developing your first application in minutes:

6. Voice Cloning

The ElevenLabs voice cloning tool lets you create your own AI voice by uploading a short recording of your voice or a voice you have permission rights to. The voice recording trial must include one speaker with no background noise and be over one minute long. You can instantly use your voice to generate speech in 29 languages and over 50 accents!

Cloning your voice with ElevenLabs AI is simple:

1. Choose between Instant or Professional voice cloning. You can also design new randomly generated voices or add a voice from the Voice Library.
2. Upload voice samples (one minute for Instant, at least 30 minutes for Professional).
3. ElevenLabs will verify your voice your's and meets quality standards.
4. Generate audio instantly with Instant voice cloning and get results after around four weeks with Professional voice cloning.

The voice clones are impressively accurate and sound indistinguishable from the original voice.

If you're uploading multiple voices, ensure the recording conditions are the same. For example, have the microphone at the same distance from the speaker without background noise. Also, keep the delivery the same by matching it with context. For example, if you want to use your voice for an audiobook, then record your voice in an audiobook style.

Whether creating a voice clone for videos, audiobooks, podcasts, video games, or chatbots, you can create your own AI voice quickly and efficiently.

7. Voice Library

The ElevenLabs Voice Library is an expanding collection of high-quality AI voices that spans a wide range of diversity. You'll never feel like there's a lack of options for finding the perfect voice for your project.

ElevenLabs AI makes finding the best voice as easy as possible. Use the filters to organize voices based on gender, age, and accent for your video, audiobook, video game, or blog. You can also add your own voices to the Voice Library using ElevenLab's Voice Design tool to get text character rewards!

Whether you're looking for a soothing narrator for your audiobook or a quirky character for your video game, the Voice Library has endless creative possibilities.

How to Use ElevenLabs Text-to-Speech

Here's how to generate realistic AI voices using ElevenLabs Text-to-Speech:

1. Create an Account
2. Select Text to Speech
3. Choose an AI Voice
4. Select Your Model
5. Insert Your Text & Generate
6. Refine Voice Settings
7. Download!

1. Create an Account

To start using ElevenLabs, I went to the ElevenLabs homepage and selected “Get Started Free.” From there, I signed up using my email.

This immediately took me to the ElevenLabs Speech Synthesis tool, where I could create lifelike speech in various languages using AI. They didn't waste any time; I didn't have to put in a credit card, and the process was straightforward and hassle-free.

I was also impressed with how simple and user-friendly the interface was. There was no need for a tutorial; everything was self-explanatory.

2. Select Text to Speech

Within the Speech Synthesis tab, I could access Text to Speech or Speech to Speech. I chose Text-to-speech.

3. Choose an AI Voice

Next, I was asked to choose my AI voice. Since I'm writing this near the holidays, it felt suitable to go with the Santa Claus voice, but there are dozens to choose from. You can also create your own AI voice through ElevenLab's VoiceLab by selecting “Add voice.”

ElevenLabs offers a wide range of AI voices in different accents and tones. The color-coded tags make it easy to find the perfect voice for any project, whether it's a professional presentation or a fun video.

4. Select Your Model

I skipped the voice settings to see how my AI voice would sound without altering it. I moved on to selecting the model I wanted to use and kept it on default (Eleven Multilingual v2) for the best quality. If you are considering using your AI voice in a project such as an app, opt for the Eleven Turbo v2 for the lowest latency.

5. Insert Your Text & Generate

Next, I inserted a short blurb from ChatGPT of what I would imagine Santa would say, but you can insert text up to 5,000 characters!

For generating audio for longer texts like audiobooks, use Projects instead. By breaking the text into shorter segments, Projects produces high-quality audio while offering advanced features such as multiple speakers.

I hit “Generate.” Within a few seconds, I created an audio trial of my text that I could hit play to preview.

The way Santa pronounced, “Ho, ho, ho!” sounded inconsistent. However, this was easily solved by making simple changes in the text punctuation.

6. Refine Voice Settings

I also adjusted some voice settings by increasing the stability to make the voice slightly monotonous. I could also enhance the clarity and style, but I kept those the same.

7. Download!

Once I was happy with it, I instantly downloaded an MP3 version of the voiceover by hitting the little obtain button on the bottom right.

Despite some minor changes I implemented to my AI voiceover, ElevenLabs did an excellent job producing an authentic, high-quality voice. The default model, Eleven Multilingual v2, delivered exceptional results regarding clarity and natural-sounding speech.

Compared to other AI Voiceover generators I've used, ElevenLabs is among the best and most lifelike at an affordable price.

3 Tips for the Perfect Voiceover

There are three main things to keep in mind for the best output:

1. Be intentional about where you place punctuation. Periods, commas, and other punctuation forms significantly impact the output's delivery.
2. Take your time finding the voice that best matches the context of your content. ElevenLabs will tell you the best context for each voice.
3. Don't overlook the voice settings; refine the stability, clarity, and style for the best output.

Top 3 ElevenLabs Alternatives

When evaluating the best text-to-speech tool for your needs, it is important to consider alternatives to ElevenLabs. Let's explore a few popular options and their features to determine which tool might best fit you.

Based on the AI voice generators I have tried, here are my top ElevenLabs alternatives.

Lovo.ai

Lovo.ai is a hyper-realistic AI voice generator capable of text-to-speech and voice cloning. It offers over 500 voices in 100 languages, significantly more than ElevenLabs, which only has over 70 different voices in 29 languages. However, they do have a continuously growing Voice Library.

Additionally, Lovo.ai has some features worth mentioning that ElevenLabs lacks. Lovo.ai has a video editor where you can access thousands of royalty-free assets. Plus, it has an AI Writer that can generate script ideas and help streamline your content creation process.

For more voice and language options, plus a video editor and AI writer, choose Lovo.ai. If you have decision paralysis and/or are a game developer looking for the perfect voices for your characters, ElevenLabs is the better choice at a more affordable price.

Read our Lovo Review or visit Lovo.

Speechify

With over 25 million listeners, Speechify is a platform that reads aloud to you, cutting your practicing time in half. This tool is invaluable for students cramming for exams, employees catching up on work emails, individuals with dyslexia or ADHD who struggle with reading, or anyone who wants to consume content hands-free.

Speechify has other valuable features like text-to-speech, an AI voice studio, and AI avatars. Plus, it's compatible with many platforms, such as an iPhone, iPad, Mac app, Android app, Chrome extension, Edge add-on, and PDF Reader.

Speechify and ElevenLabs both offer incredibly natural-sounding text-to-speech capabilities. However, if you want to read content quicker, generate videos with AI avatars, and prioritize accessibility, choose Speechify. For natural AI voices perfect for video games, narrating videos, audiobooks, and AI chatbots in 29 different languages, choose ElevenLabs.

Read our Speechify Review or visit Speechify.

Murf

Murf AI is a versatile AI voice generator that instantly turns text into speech. Whether you're an educator, marketer, author, podcaster, etc., it's perfect for any content.

Murf has many similar features to ElevenLabs (text-to-speech, API, AI dubbing and translation, and voice cloning). However, Murf AI has additional features that could be game-changers, like voice-over video and add-ons for Google Slides and Canva.

It's also worth noting that while Murf offers more voices than ElevenLabs, ElevenLabs has more language options.

If you want to compliment your voiceovers with videos, have more voices to choose from, or want to add voiceovers to your Google Slides and Canva projects, go for Murf AI. For the most realistic AI voices and slightly more language options, choose ElevenLabs.

Read our Murf Review or visit Murf.

ElevenLabs Review: Is It the Most Realistic Text-to-Speech Tool?

Compared to the most popular AI voice generator contenders on the market that I've tried, ElevenLabs has the most realistic AI voices that I've come across. The AI model can accurately reproduce human intonation and inflections, adapting its delivery according to the context, which no other model can match.

While ElevenLabs has some limitations, such as fewer voice and language options than other alternatives, this is overshadowed by the quality of its voice output. The attention to detail in capturing the nuances of human speech sets ElevenLabs apart from its competitors.

ElevenLabs is an affordable and reliable choice for realistic AI voices in various applications like video games, narration videos, audiobooks, and AI chatbots in 29 languages. It has a free plan, so why not experience it yourself by creating an account and exploring its features?

Frequently Asked Questions

Is ElevenLabs any good?

ElevenLabs stands out with its remarkable voice synthesis quality. The voices sound natural, and the intonation is lifelike.

Is ElevenLabs free?

Yes, ElevenLabs has a free plan where you can generate 10,000 characters per month in 29 languages. It's the most affordable AI voice generator on the market.

How to use ElevenLabs AI for free?

To use ElevenLabs AI for free forever, select “Get Started Free” on their website and sign up using your email. Your account will be created immediately, and you can start immediately; no credit card is required.

Who owns ElevenLabs?

ElevenLabs was founded in 2022 by childhood friends Mati Staniszewski (CTO) and Piotr Dabkowski (CEO), ex-Google and Palantir staffers.

What does ElevenLabs do?

ElevenLabs is a powerful text-to-speech tool that uses artificial intelligence and natural language processing to convert written text into lifelike audio. You can also turn your voice into an AI voice, instantly translate voice recordings, and more. It's the perfect tool for creating audiobooks, podcasts, and educational content.

Is ElevenLabs safe?

ElevenLabs is a safe text-to-speech tool. It prioritizes user privacy by not collecting or storing personal information and uses secure encryption to protect user data. It also implemented a deepfake detection tool (AI Speech Classifier) ever since it has been used for hateful comments in the voices of celebrities like Emma Watson.

Oh, Google. Will you ever get an AI product release right on the first try?

Less than a month after Google unveiled its long-rumored ChatGPT competitor Gemini to the world in a glossy demo video — only for the company to face criticism for what appeared and was ultimately confirmed to be staged interactions between the presenter and the AI — new research finds that the most powerful version of Gemini available now to consumers, Gemini Pro, falls behind OpenAI’s GPT-3.5 Turbo large language model (LLM) in terms of most tasks.

Yes, you read that correctly: Google’s brand new LLM, the one that has been in development for months at least, performs worse at most tasks than OpenAI’s older, less cutting-edge, free model. After all, ChatGPT Plus and Enterprise paying subscribers can already access and use the underlying GPT-4 and GPT-4V (the multimodal offering) LLMs regularly, and have had access to the former for the better part of this year.

That’s according to the work of a team of researchers from Carnegie Mellon University and one from an enterprise identified as BerriAI.

Their paper, “An In-depth Look at Gemini’s Language Abilities,” was published yesterday on arXiv.org, the pre peer-review and open access science site. As it states plainly near the top: “In sum, we found that across all tasks, as of this writing (December 19, 2023), Gemini’s Pro model achieved comparable but slightly inferior accuracy compared to the current version of OpenAI’s GPT 3.5 Turbo.”

For the Google researchers who have spent hard hours working on Gemini — and their leadership — that conclusion has got to sting. We reached out to Google and a spokesperson responded after this story published, maintaining Google’s own research shows Gemini Pro performs better than GPT-3.5, and that an upcoming, even more powerful version, Gemini Ultra, due out in early 2024, scored higher than GPT-4 on Google’s internal research. Here’s their response in full:

• “In our technical paper [published here], we compare Gemini Pro and Ultra to a suite of external LLMs and our previous best model PaLM 2 across a series of text-based academic benchmarks covering reasoning, practicing comprehension, STEM, and coding.
• These results [in Table 2 on Page 7 of the report] show that the performance of Gemini Pro outperforms inference-optimized models such as GPT-3.5, performs comparably with several of the most capable models available, and Gemini Ultra outperforms all current models. 
• On Gemini Ultra specifically, on MMLU, it can outperform all existing models, achieving an accuracy of 90.04%. It is also the first model to exceed this threshold, with the prior state-of-the-art result at 86.4%.

“Also, it’s worth practicing the Gemini authors discussion on the nuance of these evaluations in the paper (also on the same page), pulling it out for ease: 

‘Evaluation on these benchmarks is challenging and may be affected by data contamination. We performed an extensive leaked data analysis after training to ensure the results we report here are as scientifically sound as possible, but still found some minor issues and decided not to report results on e.g. LAMBADA (Paperno et al., 2016).

As part of the evaluation process, on a popular benchmark, HellaSwag (Zellers et al., 2019), we find that an additional hundred finetuning steps on specific website extracts corresponding to the HellaSwag training set (which were not included in Gemini pretraining set) Strengthen the validation accuracy of Gemini Pro to 89.6% and Gemini Ultra to 96.0%, when measured with 1-shot prompting (we measured GPT-4 obtained 92.3% when evaluated 1-shot via the API).

This suggests that the benchmark results are susceptible to the pretraining dataset composition. We choose to report HellaSwag decontaminated results only in a 10-shot evaluation setting. We believe there is a need for more robust and nuanced standardized evaluation benchmarks with no leaked data. So, we evaluate Gemini models on several new held-out evaluation datasets that were recently released, such as WMT23 and Math-AMC 2022-2023 problems, or internally generated from non-web sources, such as Natural2Code.

We refer the reader to the appendix for a comprehensive list of our evaluation benchmarks. Even so, model performance on these benchmarks gives us an indication of the model capabilities and where they may provide impact on real-world tasks.

For example, Gemini Ultra’s impressive reasoning and STEM competencies pave the way for advancements in LLMs within the educational domain. The ability to tackle complex mathematical and scientific concepts opens up exciting possibilities for personalized learning and intelligent tutoring systems.'”

What the researchers tested

The new paper from the CMU and BerriAI researchers goes on to note that they actually tested four different LLMs: Google Gemini Pro, OpenAI GPT-3.5 Turbo, GPT-4 Turbo, and Mixtral 8x7B, the new open-source model from well-funded French startup Mistral that took the AI community by storm last week with its sudden, unceremonious arrival — dropped as a torrent link with no documentation — and its high performance and benchmark scores (standardized evaluations of AI performance).

The researchers used an AI aggregator site, LiteLLM, over a period of 4-days, December 11-15, 2023, and ran all the models through a set of different prompts, including asking them 57 different multiple choice questions “across STEM, the humanities, the social sciences,” as part of a “knowledge-based QA” test.

In that test, “Gemini Pro achieves an accuracy lower than that of GPT 3.5 Turbo, and much lower than that of GPT 4 Turbo,” specifically a score of 64.12/60.63 (out of 100/100) compared to GPT-3.5 Turbo’s 67.75/70.07, and GPT-4 Turbo’s 80.48/78.95. See the top row of the following table included in their paper.

Interestingly, the researchers found that when prompting the different LLMs to choose between answers labeled A, B, C, or D, Gemini disproportionately chose “D” more times than the other models, irrespective of it was the right answer.

“Gemini has a very skewed label distribution, biased towards selecting the final choice of ‘D’ which contrasts to the result of the GPT model, which is more balanced,” the paper states. “This may indicate that Gemini has not been heavily instruction-tuned towards solving multiple-choice questions, which can cause models to be biased with respect to answer ordering.”

In addition, the researchers observed that Gemini was worse than GPT-3.5 Turbo on several specific categories of questions, namely, human sexuality, formal logic, elementary math, and professional medicine. The researchers stated that this was in no small part due to the fact that Gemini refused to answer some questions, stating it could not comply due to its safety and content restrictions, which the researchers counted as an erroneous response in their grading/benchmarking.

Gemini Pro did outperform GPT-3.5 Turbo in two categories of multiple choice questions — security and high school microeconomics, but “for the two tasks where Gemini Pro outperformed GPT 3.5 Turbo, gains were marginal,” the researchers stated. Also, GPT-4 still reigned king over all the models tested.

To be fair to Gemini, the researchers were careful to note it outperformed GPT-3.5 in one other case: when the output of the LLMs were greater than 900 tokens long (tokens refer to the different numeric values assigned to different words, letter combinations, and symbols, which reflects the model’s internal organization of different concepts).

The researchers tested the models on another category of questions, “general purpose reasoning,” where no answer options were presented. Instead, the LLMs were asked to read a logic problem and respond to it with what they thought was the correct answer.

Once again, the researchers found “Gemini Pro achieves an accuracy slightly lower than that of GPT 3.5 Turbo, and much lower than that of GPT 4 Turbo…Gemini Pro underperformed on longer, more complex questions while the GPT models were more robust to this. This was particularly the case for GPT 4 Turbo, which showed very little degradation even on longer questions, indicating an impressively robust ability to understand longer and more complex queries.”

Yet Gemini did manage to best “all GPT models,” including GPT-4, on two subcategories here: word sorting and symbol manipulation (Dyck language tasks). As the researchers put it: “Gemini is particularly good at word rearrangement and producing symbols in the correct order.”

When it came to math and mathematical reasoning, the researchers identified a similar result as in testing the other subject matter: “Gemini Pro achieves an accuracy slightly lower than that of GPT 3.5 Turbo, and much lower than that of GPT 4 Turbo.”

Think Gemini might redeem itself in programming? Think again. When given two different strings of incomplete Python code to complete, Gemini performed “lower than GPT 3.5 Turbo and much lower than GPT 4 Turbo on both tasks.”

And when asked to act as “web agent,” navigating the public internet and completing tasks on behalf of the user based on prompted instructions, “Gemini-Pro performs comparably but slightly worse than GPT-3.5-Turbo.”

Gemini did outshine all other models in one area that seems uniquely well suited to Google’s prior skill set: translating content between languages. As the researchers note: “Gemini Pro outperforms both GPT 3.5 Turbo and GPT 4 Turbo on 8 out of 20 languages, and achieved the top performances on 4 languages.”

But even this result was sullied by the fact that “Gemini Pro showed a strong tendency to to block responses in approximately 10 language pairs,” suggesting an overzealous content moderation/safety system in place.

What does it mean for Google’s AI ambitions and for users?

The results are clearly a blow to Google’s ambitions to go head-to-head with OpenAI in the generative AI race, and with the more powerful Google Gemini Ultra model not due out until next year, it will likely mean that Google remains behind in AI performance at least until then.

Interestingly, though, the study also showed that Mistral’s hit new LLM Mixtral 8x7B — which utilizes a “mixture of experts” approach, wherein several different smaller AI models are chained together, each handling different sets of tasks for which they are ideally specialized — also performed much worse than OpenAI’s GPT-3.5 Turbo across the board, for the most part. And Gemini Pro “outperforms Mixtral on every task that we examined,” according to the researchers.

That suggests a bright spot for Google’s AI work: it is still better than the cutting-edge open source.

Yet, overall, it is hard not to walk away from this study with the impression that OpenAI is, for now, still the king of consumer and enterprise-facing generative AI.

AI influencers such as University of Pennsylvania Wharton School of Business professor Ethan Mollick largely seem to agree. As Mollick posted on X today: “For most individual cases, you want to use the best AI & that is clearly still GPT-4…at least until Gemini Ultra is released in the new year.”