SAN FRANCISCO — After testing some of the latest in artificial intelligence, businesses and nonprofit groups agree on one thing: It’s already changing the way they operate.
Five organizations that were among the first to get access to GPT-4, the latest product from San Francisco startup OpenAI, said in interviews that they were rehiring staff, reshaping internal teams and building their infrastructure in anticipation of the technology. Re-evaluating strategies. their work.
Their experiences support the idea that, for better or worse, AI technology may very soon fundamentally change the daily lives of some people.
But the organizations also said that customizing the technology to their specific needs required a tremendous amount of work, with employees providing daily feedback to the software to adapt it to terminology and methods specific to their fields, such as education or finance. can be trained from. OpenAI, best known for creating the AI chatbot ChatGPT, can then integrate data from that work into its own models to potentially improve its technology.
In fact, each of the early testers is a microcosm of what others might do as access to GPT-4 expands.
“The perception now in the market is that you plug into these machines and they give you all the answers,” said Jeff McMillan, head of analytics, data and innovation at Morgan Stanley’s Wealth Management Division.
This is not true, he said. He added that the bank has 300 employees who are spending some of their time testing their technology using GPT-4.
“We have a team of people who literally review every response from day one,” he said.
For Morgan Stanley, the result has been a specialized chatbot built with GPT-4 that serves as an internal research tool for the financial advisor’s staff. McMillan said the tool has been trained not only on 60,000 research reports on parts of the global economy, but also on another 40,000 of the firm’s internal documents – making it an expert on any financial topic a financial The advisor wants to see.
To be sure, the early adopters of GPT-4 are not a random sample of the economy. OpenAI, which turned profitable in 2019, has picked up organizations over the past weeks and months.
Critics of OpenAI and its rivals allege that the AI field has benefited from skeptical hype over the past several months. OpenAI was looking for positive examples to show it, said founder Sal Khan, a non-profit educational organization, Khan Academy, six months ago.
“The context was: we are going to work on the next generation model; We want to be able to launch it with positive use cases,” he said.
Khan Academy is best known for its videos on YouTube, but since being approached by OpenAI, Khan said it has poured resources into creating KhanMigo, a chatbot tutor that specializes in teaching established concepts. is trained.
“We collectively spent about 100 hours fine-tuning the model so that it potentially behaves like a good teacher,” he said.
“If you look at the cost of tuition, it can be a pretty big deal,” Khan said. “It’s like having an amazing graduate student or tutor or professor that you can start talking to in an instant.”
Stripe, a tech company that makes payment software and related products for businesses, said that when it got early access to GPT-4 in January, it pulled 100 employees from their regular jobs and put them through an internal “hackathon”. assigned in which each person spent a week on an average test of ideas.
Duolingo, an app for language learning, reached GPT-4 in the fall, and employees said CEO Louis von Ahn was so impressed with it that he called a meeting at 8 a.m. the next morning and immediately changed people’s jobs.
“He said, then, ‘pivot your team,'” said Edwin Bauz, a product manager. “Since then, we have been working closely with GPT-4 and the OpenAI team.”
As of now, Duolingo has added a new, paid subscription tier costing $29.99 per month or $167.88 annually, which allows access to AA conversation chatbots in French or Spanish. They’ve also added an AI bot that will explain grammatical concepts to you as you move through specific Duolingo lessons.
According to Bodge, the company has designed 1,000-2,000 word prompts for GPT-4 that power the bots. The company will not share signals upon request.
All of the organizations NBC News spoke with said they were proceeding with some degree of caution, given that AI technology is so new and the potential risks are unknown. Mike Buckley, CEO of Be My Eyes, a company that makes an app for people who are blind or have low vision, said he’d like to get the trial version of the app with GPT-4 into more hands, “but we want thoughtful and want to be safe.
“Can we launch this more broadly to the community in six to eight weeks? It’s possible, but we’ll go where the data and use cases take us,” he said.
The company works by pairing people with low vision with volunteers who, over video calls, can tell app users what’s around them — such as product labels at the grocery store, directions through the airport, or greeting cards Wording in . The version with GPT-4 works without a volunteer on the other end as the AI describes what it “sees” with the camera.
According to a video posted on TikTok, one of the app’s blind spokespeople used it to get directions on the London Underground subway system.
“We’ve tried to break it down,” Buckley said, adding that his staff conducted thousands of tests. “We’ve spent several weeks critiquing the technology, and we’ve been pleasantly surprised.”
He said that his company had no security concerns with GPT-4, but that it had made mistakes; For example, mixing a toaster for a slow cooker on a website.
