ChatGPT turns a year old, marking a major milestone for generative AI

There have also been plenty of existential questions about everything from the future of work and the creative process to the ethics of innovation and the role of regulation. The pace of progress has even led some to fear AI’s unchecked progress could also harm humanity far beyond the realm of ChatGPT.

Amidst all the innovation, ChatGPT has also sparked new litigation and even legislation to challenge what OpenAI and other companies should be allowed to do with large language models (LLMs). Multiple lawsuits this year have alleged OpenAI’s approach to AI has violated copyright and data privacy laws, sparking new debates about what current laws protect and whether governments need to pass new rules regulating LLMs. 

To celebrate ChatGPT’s anniversary, OpenAI published a short video highlighting some of the ways people have used the platform for both work and everyday life, but the company OpenAI declined to speak with Digiday about the past year. 

That said, the traffic speaks for itself: Since it debuted on Nov. 30, 2022, ChatGPT’s mobile and desktop web traffic has skyrocketed from 153,000 visits on release day to 266 million in December 2022 to 1.7 billion in October 2023, according to Similarweb.

Last month, worldwide traffic for ChatGPT was 1.67 billion, according to Similarweb’s preliminary estimate. That’s far higher than Google Bard’s 267 million, Character.AI’s 165 million and Anthropic-owned Claude’s 26 million. In the U.S. alone, ChatGPT attracted 182 million visits last month while its app grew to around 4.9 million monthly active users across iOS and Android devices.

“On one hand, we’re at the earliest stages of figuring out what this can do,” said Brian Yamada, chief innovation officer at the WPP-owned agency VMLY&R. “If you think about the first wave of apps on the iPhone, there was the flashlight and iBeer. It’s not until we really master the media that we discover the use cases that really matter.”

ChatGPT has also been a gateway drug for more AI investments. Earlier this year, a Gartner survey of 2,500 executives found 45% of respondents cited ChatGPT’s popularity as reason to increase their AI investments. One of the major ways ChatGPT has had an impact is in making generative AI more accessible, according to Gartner analyst Nicole Greene. She pointed out that market penetration for generative AI was less than 1% two years ago and was between 5% and 20% this summer.

“You don’t think about the true impact that tech has on you until you go back and say, ‘What did I do without this before?’” Greene said. “Can you imagine a time when you didn’t search the internet on your phone? Generative AI is following the hype cycle we’ve seen around emerging technology….It’s following the hype cycle, but it’s so much faster.”

Some companies that were early to integrate with ChatGPT have already seen results. Klarna, the Swedish fintech and e-commerce company, created a plugin to help people personalize their shopping experience that now runs around 20,000 queries per week. Semantic search powered by LLMs has helped Klarna better understand consumer intent across popular categories — such as clothing, outdoor gear, and pet care — and when people search for them. For example, ChatGPT queries for electronics are popular at 7 a.m. before work, clothing shopping is consistent throughout the day and pet care searches peak in the afternoon.

“We don’t see the conversations or anything like that, but we see the searches that are made from ChatGPT today,” said Martin Elwin, Klarna’s director of engineering. “…What we see that means for the product database is to allow consumers to ask even more open questions going forward.”

Since Duolingo began testing OpenAI’s LLMs last fall, ChatGPT and GPT-4 helped the language learning app create content and quickly create new products like teaching tools for text and audio lessons. It’s even helped Duolingo with “learning about our learners and how they learn,” said Klinton Bicknell, Duolingo’s head of AI and director of AI research.

“It would have been minimally very slow, but ultimately probably just not even feasible to release a feature like that,” Bicknell said. “But now a small team did it in less than a [fiscal] quarter. That’s the added speed we get from being able to use these kinds of tools.”

Here’s a look at some of the milestones ChatGPT has had in the past year — high, low and otherwise:

• January: Microsoft’s multibillion dollar investment in OpenAI and broader partnership sparked a new flurry of interest around ChatGPT and generative AI.

• February: Just two months after ChatGPT launched, OpenAI announced a new subscription plan called ChatGPT Plus, offering users that pay $20 per month to get faster responses and early access to new features. 

• March: OpenAI debuted its new APIs for ChatGPT and Whisper — an AI model for speech recognition — along with new partners like Snap, Instacart and Shopify. A few weeks later, OpenAI announced new plugins for ChatGPT featuring a range of major brand partnering including Expedia, Kayak, Klarna, Slack, OpenTable and Zapier. Meanwhile, researchers documented more than 100 examples of ChatGPT spreading misinformation that advanced “false narratives” on various topics. 

• April: Amid more concerns about ChatGPT’s problems with misinformation and “hallucination,” OpenAI addressed its approach to AI safety. The company also pointed out that its newest model, GPT-4, was 40% more likely than GPT-3.5 to produce factual answers.

• May: OpenAI cofounder and CEO Sam Altman made his debut in Congress during the U.S. Senate’s first AI hearing, where he answered a range of question from lawmakers about the risks of AI and what he thinks about government regulation. Two days later, ChatGPT’s new iOS app debuted in Apple’s App Store, which offered a number of new features including early access to OpenAI’s GPT-4 for ChatGPT Plus subscribers. It also announced a new partnership with Nextdoor.

• June: ChatGPT experienced its first use decline as web traffic dropped 10% during an early-summer lull. The same month, concerns related to ChatGPT led to two separate lawsuits alleging OpenAI’s data collection and content creation practices violated copyright and privacy laws.

• July: OpenAI and other companies — including Anthropic, Google, Amazon and Microsoft — met with the White House to address growing concerns about AI and the companies’ plans for ethical development. The same month, OpenAI became one of the founding members of the Frontier Model Forum, a cross-company organization focused on “safe and responsible development of frontier AI models.” It also announced a new partnership with the American Journalism Project and committed more than $5 million for helping local news organizations use AI.

• August: OpenAI debuted its enterprise version of ChatGPT that offers new security and privacy protections for companies using OpenAI’s LLMs.

• September: ChatGPT added new ways to “see,” “hear” and “speak” — thanks to new multimodal capabilities — and days later announced the platform could finally browse the internet. 

• October: The text-to-image generator DALL-E 3 was added into ChatGPT Plus and ChatGPT Enterprise, giving users a new way to make images directly within the same platform. The same month, OpenAI and the other Frontier Model Forum members committed another $10 million toward AI safety days before President Joe Biden signed his executive order on AI.

• November: OpenAI hosted its first developer conference, DevDay, and also debuted new ways to easily build custom versions of ChatGPT. Later in the month, the company experienced a dramatic week after the company’s board unexpectedly fired — and then rehired — Altman and cofounder Greg Brockman. OpenAI was also hit with another lawsuit alleging ChatGPT violated U.S. copyright law when it scraped an author’s book and generated answers based on the content.