How OpenAI ushered in the generative AI boom: a brief history
Michelle Cheng discusses AI at length in “AI hallucinations: Tune in, turn on, beep boop,” episode 4 of the latest season of the Quartz Obsession podcast. Listen now or find it wherever you get your podcasts: Apple Podcasts | Spotify | Google | Stitcher | YouTube
Or, read the transcript!
It was not a tech giant like Google or Apple that created ChatGPT. Rather, the AI bot came from a little-known 300-employee startup called OpenAI, which was founded by Silicon Valley elites with Elon Musk and Sam Altman as early board members.
OpenAI’s release of ChatGPT in November 2022 took the internet by storm with its ability to generate content ranging from haikus to short stories in human-sounding prose. Within two months of launch, ChatGPT had amassed millions of users.
To be sure, the AI bot has its flaws, from confidently giving the wrong answer to making stuff up. The so-called AI hallucinations pose a big challenge to the future of the AI industry.
Still, OpenAI’s release of ChatGPT has spurred companies from Google to Bain to say that they’re incorporating generative AI into their services. Here’s a look at how OpenAI skyrocketed to being the buzziest tech company of the moment.
It starts with a lofty goal
2015: OpenAI was launched as a nonprofit by a group of Silicon Valley titans including Sam Altman, Elon Musk, Peter Thiel, and LinkedIn cofounder Reid Hoffman. The group created a research lab focused on building software that could replicate human intelligence and capabilities, better known to AI researchers as “general artificial intelligence.” Musk and others pledged $1 billion in donations, reported the New York Times.
The company’s mission was to create AI tools that “benefit all of humanity.” At first, the lab pledged to open source all of its research, as Musk and Altman said that the threat of harmful AI would be mitigated if everyone had access to the technology. But as the lab started building its AI tools, some OpenAI employees said that some of its ideas and code should be walled off from the public, according to The Times, arguing that openly sharing OpenAI’s software could exacerbate the generation and distribution of false information.
Musk says peace
2018: Musk resigned from OpenAI’s board, reportedly in part due to his growing conflict of interest with OpenAI. By then, Musk was building his own AI project at Tesla: Autopilot, the driver-assistance technology that automatically steers, accelerates, and brakes cars on highways. He even reportedly poached a key employee from OpenAI.
“There is disagreement, mistrust, egos,” Mr. Altman said in an interview with the Times. “The closer people are to being pointed in the same direction, the more contentious the disagreements are. You see this in sects and religious orders. There are bitter fights between the closest people.”
OpenAI begins a relationship with Microsoft
2019: Microsoft announced it would invest $1 billion in OpenAI, allowing the company to train and run its AI models on Azure, Microsoft’s cloud computing service. The funding marked a major turning point for OpenAI.
That same year, OpenAI moved to a “capped profit” structure, which allowed it to build massive computing power and attract top talent, as the Wall Street Journal reported.
OpenAI releases a realistic-looking art generator
2021: OpenAI introduced DALL-E, an AI system that creates images from text descriptions. The model would spur competitors like Midjourney and Stability AI.
ChatGPT goes through multiple iterations
2022: In early 2022, the team started building ChatGPT. When the chatbot was ready, OpenAI let beta users test the technology, but it wasn’t clear to users what they were supposed to talk about with the bots, Greg Brockman, an OpenAI cofounder and current president, told Fortune.
OpenAI switched gears and tried to build expert chatbots that could help professionals in specific areas, but that also didn’t work out as planned. So, instead, the company decided to release ChatGPT to the public in its current form.
ChatGPT goes live
Nov. 2022: On Nov. 30, ChatGPT launched to the world.
More companies go all in on generative AI
Jan. 2023: Microsoft rolled out the latest Bing search engine, powered by generative AI. A few weeks later, Google released Bard, the AI chatbot competitor. Big Tech’s race to dominate AI was officially underway.
In the same month, OpenAI announced a partnership with Bain. The consultancy firm said it would incorporate “OpenAI technologies into its internal knowledge management system, research, and processes to improve efficiency.” The following month, Salesforce announced it would be integrating OpenAI technologies into Slack, which it owns, with the goal of helping users send and retrieve information faster.
A better, faster version of ChatGPT is released.
March 2023: On March 14, OpenAI released ChatGPT-4, a more accurate version of its previous chatbot. The company claimed that ChatGPT-4 was able to pass the bar exam with a score in the top 10% of test takers.
The newest model accepts images. On a livestream on March 14, OpenAI’s president, Greg Brockman, demonstrated by uploading into ChatGPT a photo of a handwritten note describing what he wants a website to look like. The tool generated code that he could use to build that exact site.
Regulators quickly turned their attention to generative AI companies. Italy temporarily banned ChatGPT over concerns of the processing of personal data for its model, making it the first Western country to take action against a chatbot.
OpenAI responds to privacy concerns
April 2023: A spokesperson for OpenAI, Hannah Wong, told the New York Times that the company generated profits for investors, but that it is still governed by a nonprofit and that its profits were capped.
On April 25, OpenAI announced that users can turn off ChatGPT’s chat history, which would also disable future training on the unsaved chats.
This surely won’t be the last update for OpenAI and the rest of the rapidly developing AI industry.