The AI revolution is coming. But not as fast as some people think.
By Steve Lohr
Lori Beer, the global chief information officer of JPMorgan Chase, talks about the latest artificial intelligence with the enthusiasm of a convert. She refers to AI chatbots like ChatGPT, with its ability to produce everything from poetry to computer programs, as “transformative” and a “paradigm shift.”
But it’s not coming soon to the nation’s largest bank. JPMorgan has blocked access to ChatGPT from its computers and told its 300,000 workers not to put any bank information into the chatbot or other generative AI tools.
For now, Beer said, there are too many risks of leaking confidential data, questions about how the data is used and about the accuracy of the AI-generated answers. The bank has created a walled-off, private network to allow a few hundred data scientists and engineers to experiment with the technology. They are exploring uses such as automating and improving tech support and software development.
Across corporate America, the perspective is much the same. Generative AI, the software engine behind ChatGPT, is seen as an exciting new wave of technology. But companies in every industry are mainly trying out the technology and thinking through the economics. Widespread use of it at many companies could be years away.
Generative AI, according to forecasts, could sharply boost productivity and add trillions of dollars to the global economy. Yet the lesson of history, from steam power to the internet, is that there is a lengthy lag between the arrival of major new technology and its broad adoption — which is what transforms industries and helps fuel the economy.
Take the internet. In the 1990s, there were confident predictions that the internet and the web would disrupt the retailing, advertising and media industries. Those predictions proved to be true, but that was more than a decade later, well after the dot-com bubble had burst.
Over that time, the technology improved and costs dropped, so bottlenecks fell away. Broadband internet connections eventually became commonplace. Easy-to-use payment systems were developed. Audio and video streaming technology became far better.
Fueling the development were a flood of money and a surge of entrepreneurial trial and error.
“We’re going to see a similar gold rush this time,” said Vijay Sankaran, chief technology officer of Johnson Controls, a large supplier of building equipment, software and services. “We’ll see a lot of learning.”
The investment frenzy is well underway. In the first half of 2023, funding for generative AI startups reached $15.3 billion, nearly three times the total for all of last year, according to PitchBook, which tracks startup investments.
Corporate technology managers are sampling generative AI software from a host of suppliers and watching to see how the industry shakes out.
In November, when ChatGPT was made available to the public, it was a “Netscape moment” for generative AI, said Rob Thomas, IBM’s chief commercial officer, referring to Netscape’s introduction of the browser in 1994. “That brought the internet alive,” Thomas said. But it was just a beginning, opening a door to new business opportunities that took years to exploit.
In a recent report, the McKinsey Global Institute, the research arm of the consulting firm, included a timeline for the widespread adoption of generative AI applications. It assumed steady improvement in currently known technology, but not future breakthroughs. Its forecast for mainstream adoption was neither short nor precise, a range of eight to 27 years.
The broad range is explained by plugging in different assumptions about economic cycles, government regulation, corporate cultures and management decisions.
“We’re not modeling the laws of physics here; we’re modeling economics and societies, and people and companies,” said Michael Chui, a partner at the McKinsey Global Institute. “What happens is largely the result of human choices.”
Technology diffuses across the economy through people, who bring their skills to new industries. A few months ago, Davis Liang left an AI group at Meta to join Abridge, a health care startup that records and summarizes patient visits for physicians. Its generative AI software can save doctors from hours of typing up patient notes and billing reports.
Liang, a 29-year-old computer scientist, has been an author on scientific papers and helped build so-called large language models that animate generative AI.
His skills are in demand these days. Liang declined to say, but people with his experience and background at generative AI startups are typically paid a base salary of more than $200,000, and stock grants can potentially take the total compensation far higher.
The main appeal of Abridge, Liang said, was applying the “superpowerful tool” of AI in health care and “improving the working lives of physicians.” He was recruited by Zachary Lipton, a former research scientist in Amazon’s AI group, who is an assistant professor at Carnegie Mellon University. Lipton joined Abridge early this year as chief scientific officer.
“We’re not working on ads or something like that,” Lipton said. “There is a level of fulfillment when you’re getting thank-you letters from physicians every day.”
Saving time and streamlining work inside companies are the prime early targets for generative AI in most businesses. New products and services will come later.
This year, JPMorgan trademarked IndexGPT as a possible name for a generative AI-driven investment advisory product.
“That’s something we will look at and continue to assess over time,” said Beer, the bank’s tech leader. “But it’s not close to launching yet.”