What to know about OpenAI, the company behind ChatGPT
The viral chatbot’s creator is rocketing into the mainstream.
A popular tool that can respond to questions in eerily human ways, called ChatGPT, has captured the internet’s attention as people use it write song lyrics, essays, TV episodes and more.
Now, OpenAI, the company behind the chatbot, is rocketing into the mainstream. Microsoft is reportedly investing up to $10 billion in the company, hoping it can use ChatGPT to resuscitate its search engine, Bing, and improve products in its Microsoft Office suite.
But OpenAI faces steep challenges, notably fixing its products’ glaring issues with accuracy, bias and harm.
Here’s everything you need to know about OpenAI.
What is OpenAI’s history, and how was Elon Musk involved?
The San Francisco-based artificial intelligence lab started in 2015 as a nonprofit, trying to build “artificial general intelligence,” or AGI, which is essentially software that’s as smart as humans.
The company wanted to protect against a future in which big tech companies, like Google, mastered AI technology and monopolized its benefits. The nonprofit’s goal was to build AI software transparently and make its products open-source so the world could benefit.
Silicon Valley notables pledged $1 billion to start it up. Donors included Tesla and SpaceX founder Elon Musk; venture capitalist Peter Thiel; and Sam Altman, who became the CEO of OpenAI in 2019. Musk left the company’s board in 2018, citing time demands of running Tesla and SpaceX.
OpenAI charted a complicated path. It would use extraordinarily large amounts of data and powerful neural networks, which is software loosely based on neurons in the human brain, to create its AI products. But the computing power and compensation costs to pull that off – one early-era OpenAI employee was paid $1.9 million in salary, according to its 2016 tax records – made it difficult to run the company as a nonprofit.
In 2019, OpenAI transitioned into a for-profit company, with an unusual structure to cap investor profits at a certain multiple of their investment. Altman also took $1 billion in funding from Microsoft, which agreed to license and commercialize some of OpenAI’s technology.
Microsoft declined to comment on its partnership with OpenAI beyond what’s public. OpenAI did not respond to a request for comment.
What does OpenAI make and who can use it?
OpenAI started by trying to build a system that understood language, taking advantage of the troves of text on the internet to learn from, OpenAI officials told The Washington Post.
In 2020, it released GPT-3, a text-generating tool that could produce plausible-sounding passages of text on demand.
After that, OpenAI tried to replicate GPT-3’s success by feeding it computer code and creating a tool called Codex, which helps computer programmers write code faster. Codex fuels GitHub’s Co-Pilot, a publicly available tool that translates human instructions into computer code for a monthly fee. (Microsoft owns GitHub.)
OpenAI also tried to combine vision with language, and trained GPT-3 to find patterns between words and images by ingesting massive data sets filled with pictures and captions from the internet. That resulted in DALL-E, which released in January 2021, and could create images based off human prompts.
Soon after, it created DALL-E 2, a program that generated even better photorealistic images.
DALL-E 2 went viral after it publicly released last year. People could enter nonsensical prompts, such as asking for a photo of a Dachsund puppy in space in the style of painted glass, and received high-quality images. The public can use DALL-E 2 for a fee. Companies can also incorporate the technology into their own apps for a cost.
In November, OpenAI released ChatGPT. The chatbot, essentially a fine-tuned version of its earlier text-generators, impressed the public with its humanlike prose. The chatbot could talk about religion, write essays and poetry or complete computer code. It also got basic facts wrong, provided racist and sexist responses and prompted worries about cheating in school. ChatGPT is estimated to have reached roughly 100 million active users in January, reports show.
On Wednesday, OpenAI said it will offer a premium version of the service, called ChatGPT Plus, for $20 per month. There will still be a free version that people can use during off-peak times.
Why are people excited about ChatGPT, and what does Silicon Valley think?
For the general public, the release of ChatGPT felt like a sudden leap forward, specifically in the field of generative artificial intelligence, where software creates content like texts or images based on descriptions.
Some industry analysts said this would spell the end for professionals such as journalists and screenwriters, though high-profile disasters incorporating the tool into news writing have cast doubt on that view.
The advances, as people in Silicon Valley have pointed out, are not exactly new. Tech titans such as Meta and Google had been working on similar technology, offering limited releases for some, and taking others down when they exhibit problematic behavior.
But with OpenAI’s strategy of releasing ChatGPT for millions to use, despite the harms it could cause, the general public got to interact with AI software in a very tangible way.
The intense interest in ChatGPT has ignited a race at companies like Google and Meta to fast-track their own AI products to the public, current and former officials from those companies earlier told The Post. Google wants to speed up its processes for making these products public, according to a report in the New York Times.
Who are the big players in AI right now?
Many tech companies are involved in artificial intelligence. Google pioneered advances in generative artificial intelligence, some of which underpin ChatGPT, and created the language model LaMDA, which a former Google engineer claimed was sentient.
The start-up Stable Diffusion launched its own version of OpenAI’s DALL-E, with fewer restrictions on how it’s used. Research lab Midjourney released another text-to-image generator in the summer, which created the illustration that sparked a controversy in August when it won an art competition at the Colorado State Fair.
In November, Meta released an AI tool called Galactica. The company pulled it down three days later after it was criticized for being inaccurate. Months before, it released a chatbot called BlenderBot 3, which reportedly made racist comments. In 2016, Microsoft created a chatbot, Tay, which it took down after a day after it was also revealed to be racist.
Does Microsoft own OpenAI?
No. Microsoft is an investor in OpenAI.
The Redmond, Wash.-based company has been chasing Google’s advances in artificial intelligence and is now pouring billions into OpenAI, in hopes that the investment can help Microsoft leapfrog its competitors.
Microsoft wants to use ChatGPT’s technology to revitalize its products, potentially having AI help create Excel spreadsheets, generate art for Power Point slides or draft an email in Outlook.
News reports indicate that Microsoft will incorporate a newer version of ChatGPT, called GPT-4, into its search engine, Bing, in a bid to overtake Google’s dominance.
– – –
The Washington Post’s Nitasha Tiku contributed to this report.
This discussion has ended. Please join elsewhere on Boston.com