In 2017, Tencent — the Chinese tech and gaming conglomerate — released a report on the AI talent shortage which claimed that there were only 300,000 AI researchers and practitioners worldwide. These 300,000 people had access to millions of job openings, which made competition for top talent so steep that many countries began poaching international talent. Today, it’s been nearly two years since this AI talent crisis first emerged, and NewtonX decided to analyze whether or not anything has changed.
To size the gap between talent and open positions in the AI sector, NewtonX surveyed 200 AI employees who graduated from college within the past year, 25 leaders in the AI education space (including new programs such as MITs computing program and Mount Sinai’s new Machine Learning master’s program), and 200 employers/recruiters at AI companies with over ten open AI positions. The data and insights in this article are sourced from this AI talent survey.
High Salaries, Steep Competition, and Competitive Poaching: Why the AI Talent Shortage is Still a Problem
AI-related job postings more than doubled from 2017 to 2018, while searches for AI jobs (ostensibly a measure of job seeker volume) increased by only 15%. The gap between the two is huge: searches per week account account for less than 25% of new postings per week.
This gap has created a highly competitive market, wherein AI experts command salaries more than 5 times higher than what professionals in the space made five years ago. In fact, multiple employers said that candidates with 3-5 years of experience commanded base pays between $300k and $500k, while those with 10+ years command salaries in the millions. Meanwhile, candidates who are close to graduating or have recently graduated are aggressively recruited by top tech companies.
This demand has spurred numerous new computer science/AI undergraduate and postgraduate programs. For instance, last year Carnegie Mellon started offering the first US undergraduate degree in AI. Applicants to Berkeley’s doctoral degree in electrical engineering and computer science surged from 300 ten years ago to 2,700 last year, with over 50% of applicants stating they’re interested exclusively in AI. Similarly, the Icahn School of Medicine at Mount Sinai recently unveiled its new Master’s Degree in Biomedical Data Science, which will include instruction on machine learning, big data analysis, and computer sciences. Meanwhile, applicants to UC Berkeley’s doctoral program in computer science and electrical engineering rose over the past ten years from 300 to 2,700, with over 50% of applicants this past year stating that they’re interested in pursuing AI. The school has since tripled its entering class size.
Despite these new programs, however, we’re still years out from the graduates of the programs entering the workforce. This has created barriers for smaller companies that can’t afford to beat out competitive salaries — sparking fears that the talent shortage could stymie AI progress.
Will American AI Progress Slow Due to Talent Shortages?
The fight for talent has become particularly competitive between the US and China. Last year, Google set up an AI center in Beijing to recruit AI students in China. Likewise, both Tencent and Alibaba have attended AI conferences to recruit students with specialized AI knowledge. China has also invested in institutions of higher learning in order to attract scientists from other countries and attract top international students. For instance, Tianjin University recently hired an American scientist to head its undergraduate program in pharmaceutical science and technology. The program is taught in English, is designed to be recognized worldwide, and is completely free — making it highly competitive and attractive to students.
Competition for AI talent will not slow for at least another three years. Universities, private organizations, and even governments will continue to compete for talent, and candidates will yield massived bargaining power. However, as AI programs multiply and applicants to these programs increase, the gap between open positions and number of candidates will start to close. NewtonX experts predict that the first boom of candidates will occur in three years, and that the gap will begin to close in 5-10 years.