This doesn’t reflect Altman’s leadership, but a broader trend in the tech industry that OpenAI itself has accelerated. Since ChatGPT launched in November 2022, the AI job market has been transformed. Zeki Research, a market research company, estimates that about 20,000 companies in the West employ experts in artificial intelligence. The rapid advances in machine learning and the potential for “platforming” – technical parlance for creating an entirely modern layer of technology – have changed the types of skills employers require and where those who possess them go. The result is a market in which AI talent, previously housed by tech giants, is becoming increasingly dispersed.
Start with skills. Tech giants like Microsoft and Google may be laying off non-engineers, but they’re looking for star researchers who can understand and build cutting-edge models. This group consists of perhaps several hundred people, such as Sutskever and Jeff Dean, who leads Google’s work on artificial intelligence. Companies covet such superstars because they can introduce breakthrough solutions that, for example, dramatically enhance the efficiency of an artificial intelligence system or reduce its susceptibility to confabulation. This makes them extremely valuable; many demand seven-figure salary packages.
Some are hired without interviews or as entire teams. In March, Microsoft hired most of the employees of Inflection ai, a startup that creates cutting-edge models, including its co-founder Mustafa Suleyman. The move reportedly attracted the attention of confidence busters at the Federal Trade Commission. (Mr. Suleyman sits on the board of The Economist’s parent company.) Mark Zuckerberg, head of Meta (Facebook’s parent company), personally emailed some researchers at DeepMind, Google’s artificial intelligence lab, trying to recruit them.
What’s more intriguing is how generative AI has changed the talent market lower down the ladder. According to data from the job site Indeed, one in 40 open developer positions in America list skills related to the “generative” artificial intelligence that makes ChatGPT so human-like. That’s a more than 100-fold enhance from its 2023 start. Amit Bhatia, co-founder of research firm Datapeople.io, says that before ChatGPT, a mid-sized tech company might employ a handful of AI engineers who built petite models for tasks such as customer sentiment analysis emails. Today, generative models can do a much better job than petite, internal efforts.
As a result, some AI engineers are now tasked with figuring out which AI system to employ and how to connect it to company data. Bhatia notes that the percentage of software engineering job postings that cite such “MLops” (brief for machine learning operations) has doubled since the beginning of 2022.
Different types of skills are also in demand. Kelsey Szot, co-founder of Adept, another AI startup, points to people who quickly learn to employ AI tools and can combine them to build something modern and impressive. Unlike unyielding PhD students, they come up with ideas that are often non-academic. But – says Mrs. Szot – they will solve the problem within a tight deadline. In the ultra-competitive world of AI startups, this is invaluable.
As a result of all this demand, the talent flow is changing. Over the years, engineers have flocked to the gigantic tech quintet: Alphabet (Google’s corporate parent), Amazon, Apple, Meta and Microsoft. Research firm Live Data Technologies tracks job shifts between companies. Of the AI workers in its database, the Gigantic Five’s cumulative net gains (hires minus exits) averaged 168 per month between January 2019 and November 2022, when ChatGPT was released. Many of those who left one of the Gigantic Five simply joined another.
However, over the next nine months, the net inflow of AI workers to the giants turned into an average monthly outflow. The giants are now increasing their AI wages again, for example by stealing fools from less huge tech companies with less impressive AI pedigrees like IBM and Oracle. However, net inflows have still not returned to their long-term average.
Where does AI talent go instead? One popular destination is Nvidia, the chipmaker whose “graphics processing units” are powering the artificial intelligence boom and whose ambitions extend beyond hardware to software and applications. Its market value topped $3 trillion this month, overtaking Apple and coming within striking distance of Microsoft, now the world’s most valuable company. Others joined more mature startups such as Databricks, a database and artificial intelligence company, and OpenAI.
However, every seventh leaver of the big-tech industry went to startups in “hidden” mode, which did not present products or announce plans. All eight authors of the article “Attention is all you need” published in 2017, which provided the algorithmic foundations of today’s generative artificial intelligence, left Google, where they worked at the time. Seven started their own companies (the second one joined OpenAI).
One of the motivations for moving to a smaller startup may be financial issues. For an AI wizard, the potential benefits of owning shares in a successful company could easily outweigh the salary and stock options offered by the tech giant. Scientists also increasingly want to work on significant problems. According to Zeki, since 2015, the number of them joining the healthcare industry every year has increased 20-fold (which may explain why Google is working on Med-PaLM 2, an AI doctor). Another motive is autonomy. “At huge companies, the brand risk is simply too great to ever release something fun,” Noam Shazeer, one of the paper’s authors, said at a venture capital conference last September. He then co-founded Character.ai, which allows users to create chatbots with different personalities.
The good news for both huge tech companies and petite startups is that the supply of an AI-powered workforce is growing. One source is academia. According to a 2011 report by Stanford University, about 41% of postdocs working in artificial intelligence went to work in industry, about the same as postdocs going to work in academia. By 2022, this figure was 71% for industry compared to 20% for academia. Universities are also teaching more artificial intelligence. Since 2017, the number of English-language AI-related degree programs has tripled. “All computer science departments are becoming machine learning departments,” says Naveen Rao of Databricks.
For U.S. companies that dominate the global artificial intelligence industry, hiring workers in other countries is another way to ease the talent shortage. In October, President Joe Biden signed an executive order aimed at relaxing immigration rules to allow more artificial intelligence experts to study and work in America. Google and Microsoft wrote to the Department of Labor to express their support for the plan. Other governments want the same. The EU plans training programs and subsidies. The Chinese government plans to attract talent to its shores, including: by establishing AI academies in Beijing and Shanghai. At every level, competition for AI workers is heating up.
© 2024, The Economist Newspaper Restricted. All rights reserved. From The Economist, published under license. Original content can be found at www.economist.com
