When ride-hailing company Uber sought to build a new factory in Pittsburgh in 2015 focused on self-driving cars, watched researchers and scientists from the nearby Carnegie Mellon University Robotics Center. Shortly thereafter, the company lured 40 employees of the center, including the director, with doubled salaries and hundreds of thousands of bonuses.
High-profile stories like these have contributed to the prevailing narrative that AI experts are leaving academia for industry in droves. But the shortage of artificial intelligence teachers at US universities is not the result of a distorted labor market, according to the report issued this month by the Center for Security and Emerging Technologies. AI experts are likely still interested in academic careers, but the recruitment of AI faculty to universities is not keeping pace with student demand. While big tech has stepped in to fill the gap, some experts urge caution, given that the incentive structure in industry differs from that in academia.
Historically, academic institutions have provided a steady stream of developers, engineers, and entrepreneurs that have fueled the AI innovation ecosystem. This stream was directly related to the teaching potential of the Faculty of Artificial Intelligence. But while student enrollment in computer science programs has skyrocketed over the past decade, universities have not hired enough computer science faculty to meet the demand. (The researchers used student demand for computer science as a proxy for student demand for artificial intelligence, since the latter is difficult to quantify.)
According to Jack Corrigan, one of the report’s authors, in response to increased student demand, universities have limited access to AI programs by restricting enrollment in high-demand classes, reducing small class sizes and tightening computer science admissions requirements. . At the same time, the number of PhDs in computer science is growing. recipients have expressed an interest in academic careers, but universities have not responded with a corresponding increase in faculty. Contrary to the popular narrative that industry poaching of computer science faculty is rife, universities are generally successful when they seek to hire AI faculty.
Tech companies, for their part, have stepped in to meet some of the student demand by offering alternative routes to AI education and training.
“Technology companies are becoming the new ‘boilers of innovation’ and functional universities where innovation happens,” said John Nosta, a member of Google’s Health Advisory Board and a founding member of the World Health Organization’s Digital Health Expert List. Nosta noted that Google, for example, does not always require a bachelor of science degree from prospective employees. “The excitement is no longer coming from universities, but from innovative transformational companies like SpaceX, Amazon, Apple, OpenAI and other for-profit businesses that are leaving the traditional model of education in the dust,” Nosta said. In his opinion, this trend is particularly noteworthy in the field of artificial intelligence.
Google isn’t the only company to respond to the AI workforce shortage by removing the college education requirement for some positions. IBM and Apple also waived that requirement, an effort they also hope can diversify the talent pool by providing access to those who didn’t have as many opportunities in early life.
“The jobs are there, and there’s one structural barrier we can remove,” former IBM CEO Virginia Rometty told The Wall Street Journal last year. She said that instead of a degree, IBM checks for “learning aptitude” and provides training. The results? “New hires”—her term for those without a four-year degree—worked at a level that was equal to or better than their college-educated counterparts.
Not everyone agrees that industry is the right place to train AI experts.
“I wouldn’t want to call technology companies ‘universities,'” said Dan Rockmore, a professor of computer science at Dartmouth College. “They’re really only interested in a specific set of skills — hardly universal — but I really think they’re becoming a new kind of focused technical school.”
Rockmore agreed that university AI curricula don’t always match the needs of the market, but cautioned against relying on tech companies to teach AI.
“It will be a highly skilled group of workers whose output will have a tremendous impact on the way we interact and manage,” he said. “They create technocratic ‘solutions’ without a broad education perspective or perhaps a broad and thoughtful view of the implications of the work.”
Others point to the historical interaction between higher education and the technology industry. Consider, for example, the field of computer graphics, suggested Cheri Blinn, professor of computer science at Oregon State University and past president of the Association for Computing Machinery (ACM).
“Currently, the vast majority of graphics practitioners are non-specialists, although their work still depends on a core of specialists from universities and companies who continue to move the field forward.” Similarly, she noted that artificial intelligence and, in particular, machine learning — a highly sought-after subfield of the AI industry — is no different. Scientists have worked for decades to achieve cost-effective machine learning. “Now everyone is shouting about this specialty, but in reality we need people who can enter [machine learning] in practical terms, – said Blinn.
She noted that this requires a different kind of education — focused on safe use. “Universities now recognize this need, but it takes time to create new curricula,” she said.
Some were disappointed by the slow pace of science. For example, The Seattle Times editorial board is written article last month, in which the editors complained that colleges and universities in Washington, where Microsoft is headquartered, “are awarding computer-related degrees at less than half the rate that the state’s tech companies are adding new positions, not let alone fill vacancies for existing jobs.’ They blamed not students who are interested in undergraduate programs in computer science and technology careers, but the lack of available training programs at public universities. More than 7500 New freshmen at the University of Washington have applied to the School of Computer Science and Engineering, but enrollment restrictions likely mean only 550 new students will enroll in a given year, they noted.
Training the next generation of AI scientists in the US and beyond is also important to Jim Handler, director of Rensselaer Polytechnic’s Institute for Data, Artificial Intelligence and Computation and chair of ACM’s Global Technology Policy Council.
“ACM sees this as a problem not only in the US, but around the world,” he said. “Our curriculum committees look beyond college education, but whether some of those gaps can be filled by new K-12 programs and especially high school/pre-college education programs.”
While Corrigan acknowledged that industry has a role to play in developing AI talent, he advises policymakers and academic leaders to think critically about the role of universities.
“The incentive structure for private companies is very different from the incentive structure for universities … If we want to develop a technical workforce in a fair, just, and socially optimal way, we must consider the incentives that drive the behavior of each of these actors. .”