ChatGPT’s 1-year anniversary: how it changed the world

Until one year ago, there was no ChatGPT. No clean, simple interface to write a “prompt” — that is, a question, a request, an example, a snippet — and get a detailed response in seconds. No discussions with family, friends and colleagues about “that new AI,” the one that kids are using to do homework, the one that “hallucinates,” the one making all the headlines, the one that is going to change the world/improve the world/destroy the world. No hype rocket sending the letters GPT — generative pre-trained transformer — into the stratosphere. 

Now, as we prepare to mark the day that OpenAI launched ChatGPT, I’m taking a look back at 12 months that changed the world (for better or worse) — including the Game of Thrones-like drama at the company that marked the last couple of weeks — through the lens of VentureBeat’s coverage all year long. 

On November 30, 2022, GPT-4 rumors were flying around the NeurIPS machine learning conference in New Orleans, including whispers that details about GPT-4 will be revealed there. Instead, OpenAI announced a “new model in the GPT-3 family of AI-powered large language models, text-davinci-003, part of what it calls the ‘GPT-3.5 series,’ that reportedly improves on its predecessors by handling more complex instructions and producing higher-quality, longer-form content.”

At the same time, almost as an afterthought, the company launched what it called “an early demo” of ChatGPT, another part of the GPT-3.5 series in an “interactive, conversational model” whose dialogue format “makes it possible for ChatGPT to answer followup questions, admit its mistakes, challenge incorrect premises, and reject inappropriate requests.” ChatGPT quickly became the fastest-growing consumer application in history.

Two weeks after ChatGPT’s release, everyone was talking about it — but the massive hype was accompanied by a drumbeat of criticism. The tool was described as everything from a “sensation” and “the most disruptive technology since [fill in the blank]” to a “world-class bull**** artist” and “kind of like that drunk guy or gal you meet at the bar who never stops talking, blathers on and on with an engaging combination of facts and random bullshit, but that you’d certainly never want to take home to your parents.”

OpenAI CEO Sam Altman suddenly jumped into the Twitter fray (it was still Twitter back then) with a note of caution: “ChatGPT is incredibly limited, but good enough at some things to create a misleading impression of greatness, he tweeted. “It’s a mistake to be relying on it for anything important right now. It’s a preview of progress; we have lots of work to do on robustness and truthfulness.” 

By January, there was already fierce debate about how businesses, organizations and institutions would respond to the rise of large language models that can help communicate — or borrow, or expand on, or plagiarize, depending on your point of view — ideas.

A machine learning conference debating the use of machine learning turned out to be an early example: In its call for paper submissions, the International Conference on Machine Learning noted that “papers that include text generated from a large-scale language model (LLM) such as ChatGPT are prohibited unless the produced text is presented as a part of the paper’s experimental analysis.”

The race to compete with ChatGPT was in full swing by February, when Google announced it had invested $300 million in one of the most OpenAI’s buzziest rivals, Anthropic, which had recently released its own generative AI model Claude. According to Financial Times reporting, Google would take a stake of around 10% and value Anthropic at around $5 billion.

Anthropic, VentureBeat noted at the time, was founded in 2021 by several researchers who left OpenAI, and gained more attention in April 2022 when, after less than a year in existence, it suddenly announced a whopping $580 million in funding. Most of that money, it turns out, came from Sam Bankman-Fried and the folks at FTX, the now-bankrupt cryptocurrency platform accused of fraud.

In a surprise announcement just four months after ChatGPT launched, OpenAI released the long-awaited GPT-4 model, an update of the technology behind its popular chatbot, ChatGPT. The company called GPT-4 its “most advanced system, producing safer and more useful responses.”  

GPT-4 advanced the core technology of ChatGPT by enabling the chat software to solve more difficult problems with greater accuracy, thanks to its broader general knowledge and problem solving abilities. It also added new capabilities such as accepting images as inputs and generating captions, classifications, and analyses. GPT-4 was also capable of handling over 25,000 words of text, allowing for use cases like long-form content creation, extended conversations, and document search and analysis.

It was a beautiful spring for OpenAI, and CEO Sam Altman took full advantage by soft-launching a global spring tour with an in-person meeting in April with Japan’s prime minister, during which he announced possible plans to open an OpenAI office and expand services in the country. Altman announced plans for a 17-city trek to promote OpenAI — including stops in Toronto, Washington D.C., Rio De Janeiro, Lagos, Madrid, Brussels, Munich, London, Paris, Tel Aviv, Dubai, New Delhi, Singapore, Jakarta, Seoul, Tokyo and Melbourne. 

But the tour also came at a time when OpenAI is being called out on several other fronts. There was the publication of a contentious open letter calling for an AI ‘pause,’ signed by Elon Musk, Steve Wozniak and several thousand others. There was Italy’s announcement that it would ban OpenAI’s ChatGPT due to data privacy concerns; a complaint that GPT-4 violates FTC rules; and a ChatGPT bug that exposed security vulnerabilities.

The good times at OpenAI continued in May with the announcement in a release note that it would roll out ChatGPT Plugins to ChatGPT Plus subscribers. The company said the beta release allows ChatGPT to “access the internet and to use 70+ third-party plugins.” The note said that ChatGPT Plus users would “enjoy early access to experimental new features, which may change during development. We’ll be making these features accessible via a new beta panel in your settings, which is rolling out to all Plus users over the course of the next week.”

In March, OpenAI had officially announced 11 third-party plugins, including branded offerings from Instacart, Kayak and Zapier. But the more than 70 third-party plugins now made available included ones for chess play, recipe-finding, live soccer and nutrition. For those developers who wanted to create a ChatGPT plugin, there was a waitlist for access.

By summertime, the LLM competition was heating up once again. A couple of months after Google Brain and DeepMind joined forces as Google DeepMind, with plans to take on the competitive threat posed by OpenAI and its game-changing ChatGPT, DeepMind CEO Demis Hassabis said the company was working on a new system, Gemini — which was teased at Google I/O in May — that will “tap techniques that helped AlphaGo defeat a Go champion in 2016.”

According to the report, Hassabis said the Gemini system would combine LLM technology with reinforcement learning techniques used in AlphaGo, with a goal of giving it new planning and problem-solving capabilities. But by November, the Gemini project was reportedly delayed.

Summer found OpenAI facing more challenges: The Federal Trade Commission investigated the generative AI leader for possible violations of consumer protection law. And comedian and author Sarah Silverman sued OpenAI and Meta for copyright infringement of her humorous memoir, The Bedwetter: Stories of Courage, Redemption, and Pee, published in 2010.

But it was a report that OpenAI’s GPT-4 model, which powers ChatGPT, had become “lazier and dumber” due to a “radical redesign” that prompted a response from the company’s product team.In response, Peter Welinder, VP of product at OpenAI, tweeted that not only had the company not made GPT-4 dumber, but each new version was smarter than the one before. His current hypothesis, he said, was that “when you use it more heavily, you start noticing issues you didn’t see before.” He continued: “If you have examples where you believe it’s regressed, please reply to this thread and we’ll investigate.”

OpenAI’s efforts to target the enterprise had been long anticipated, but by the time it launched ChatGPT for Enterprise in August, some wondered if the company was playing catchup.

After all, there were many other companies targeting the same enterprise business audience with generative AI — Cohere offered bespoke Large Language Model (LLM) options for the enterprise; Anthropic partnered with Scale AI to target the enterprise; and even Microsoft Azure had its own OpenAI service — but open source players were in the mix as well. Meta’s LLaMA 2, for instance, was available for commercial use.

One of ChatGPT’s biggest weaknesses from the start was its knowledge limitations — which only included information up to September 2021. But in September, OpenAI announced that ChatGPT “can now browse the internet to provide you with current and authoritative information, complete with direct links to sources,” thanks to an integration with Microsoft’s Bing search engine.

This seemed like a rerun of OpenAI’s plugin news — when OpenAI debuted ChatGPT third-party plugins, it also announced two of its own plugins — Code Interpreter (which has since been renamed “Advanced Data Analysis” and allows ChatGPT to accept uploaded files), and “Browsing” which used the Microsoft Bing API and a text-based browser to search the web and summarize information for users. But the return of web browsing of public, non-paywalled sites was heralded by the company’s leadership on their personal X accounts, with CEO Sam Altman posting “we are so back,” and CTO Mira Murati echoing the sentiment.

In October, OpenAI rolled out access to its newest image generating AI model, DALL-E 3, to users of its ChatGPT Plus subscription service and ChatGPT for Enterprise.

In a blog post announcing the news, OpenAI writes, “compared to its predecessor, DALL-E 3 generates images that are not only more visually striking but also crisper in detail. DALL·E 3 can reliably render intricate details, including text, hands, and faces. Additionally, it is particularly good in responding to extensive, detailed prompts, and it can support both landscape and portrait aspect ratios.”

November 2023

For OpenAI, and therefore ChatGPT, November began on a high, with a wave of announcements during the company’s first developer conference, Dev Day — custom GPTs! New GPT-4 Turbo! Assistants API! — crashed over Silicon Valley and the world today like a massive wave of hype and excitement.

By November 13, there was little idea of what was about to come when VentureBeat published a story that dug into the fact that according to OpenAI, the six members of its nonprofit board of directors would determine when the company has “attained AGI” —  which it defines as “a highly autonomous system that outperforms humans at most economically valuable work.” According to the company, thanks to a for-profit arm that is “legally bound to pursue the Nonprofit’s mission,” once the board decides AGI, or artificial general intelligence, has been reached, such a system will be “excluded from IP licenses and other commercial terms with Microsoft, which only apply to pre-AGI technology.” 

On November 17, OpenAI fired CEO Sam Altman — less than two weeks before ChatGPT’s first anniversary. The drama that ensued made it clear that the hype around OpenAI would likely never be quite the same, but there’s no doubt that ChatGPT’s charge into the mainstream changed the world over the past year. Now, we’ll have to wait to see what 2024 has in store for the chatbot, that, it should be clear, is “a large language model.”