AI Skills – 150 000 AI-related job postings in the US in 2018

Artificial Intelligence (AI) is reshaping economies and societies. While often in their infancy and with much potential to be fulfilled, AI technologies promise to overhaul production by improving efficiency, reducing costs, multiplying product and service offerings, and supporting decision-making, among others.

As AI permeates economies and societies, it nevertheless raises questions and fuels anxieties, including around its impact on jobs. AI is expected to complement humans in some tasks, while replacing them in others; and to generate new types of jobs while transforming the way work is organised, the tasks to be performed and the skills needed in the world of work and, more generally, in life.

Managing the AI transition in a responsible and people-centred fashion is a major challenge for governments. It calls for evidence informing the design of policies able to foster the development of AI while making sure that its deployment across economies and societies contributes to improve individual and societal well-being.
This report offers first-time evidence about the job adverts requiring Artificial Intelligence- related competences, here called AI-related jobs, and the type of AI-related competences and skills demanded from workers.

It aims to inform the discussion on skills demand and the role that human capital may play in enabling the digital transformation of firms and industries – especially the diffusion of AI-related technologies -, and to provide evidence in support of policy-making regarding technology adoption and development, economic performance and skills needs.

The analysis relies on information from online job platforms and companies’ websites collected by Burning Glass Technologies (BGT) for Canada, Singapore, the United Kingdom and the United States for the period 2012-18. It builds on the findings and AI- related keywords identified in WPIA work assessing AI-related developments in science and technology (Baruffaldi et al. (2020[1])). The analysis has further benefitted from expert advice from the UK Department for Business, Energy and Industrial Strategy (BEIS) about software packages and software repositories used in AI-related developments.

Key findings are:

• The total number of AI-related jobs advertised online increased over time, and reached almost 150 000 AI-related job postings in the US in 2018.
• In all countries, a growing number of AI-related jobs advertised online required multiple AI-related skills. In 2012, no AI-related job posted online required more than 7 AI-related skills in Canada and Singapore, or more that 9 AI-related skills in the UK and the US. In 2015 and 2018, online job adverts requiring 10 or more AI-related skills were observed in all countries.
• In all countries considered, between 2012 and 2018, the average share of AI software-related skills out of total AI-related skills sought in jobs advertised online increased. In 2018, such share amounted to about 30%.
• In 2012, a considerable part of the overall skill set of AI-related jobs advertised online was represented by skills not directly or solely related to AI, such as software engineering and development as well as operating systems. By 2018, however, software engineering and development seemed to have lost relative importance, while AI-related skills such as Natural Language Processing (NLP) and deep learning emerged more prominently. Skills related to big data constituted a considerable part of the skills profiles of AI-related jobs advertised online, throughout the period considered, in all countries.
• Skills related to communication, problem solving, creativity and teamwork gained relative importance over time and complemented software-related skills as well as AI-specific competencies, although to different extents in different countries. Overall, a trend towards requiring a set of generic skills in AI-related jobs emerges from the data.
• Between 2012 and 2016, demand for “cluster analysis” skills experienced a sudden and marked increase (a “burst”) across all countries. Moreover, particularly in Canada, the UK and the US, many of the other skills that burst in this period related to data mining and classification, NLP and computational linguistics. In the same period, across all countries evidence showed bursting of skills related to machine vision, including image recognition and processing, pattern recognition as well as motion planning.
• The burst behaviours observed in Canada and the US, and to a lesser extent in the UK and Singapore, further confirm a growing trend in the demand for deep learning-related skills, most notably in 2017 and 2018.
• In terms of occupational groups under which AI-related jobs are advertised, many AI-related jobs belonged to the “Professionals” and “Technicians and associated professionals” categories.
• Moreover, with the exception of “Skilled agricultural, forestry and fishery workers”, which nevertheless are insufficiently represented in BGT data (see Cammeraat and Squicciarini (2021, forthcoming[2]) ), a growing number of AI- related jobs openings was posted online over time, in relation to all occupational groups.
• Skills related to AI appear to be in demand across almost all sectors of the economy, though to varying degrees. The group of sectors “Information and Communication” (J), “Financial and Insurance Activities” (K) and “Professional, Scientific and Technical Activities” (M) ranked at the top in terms of AI-job intensive sectors, in all countries considered.

learning emerged more prominently. Skills related to big data constituted a considerable part of the skills profiles of AI-related jobs advertised online, throughout the period considered, in all countries.
• Skills related to communication, problem solving, creativity and teamwork gained relative importance over time and complemented software-related skills as well as AI-specific competencies, although to different extents in different countries. Overall, a trend towards requiring a set of generic skills in AI-related jobs emerges from the data.
• Between 2012 and 2016, demand for “cluster analysis” skills experienced a sudden and marked increase (a “burst”) across all countries. Moreover, particularly in Canada, the UK and the US, many of the other skills that burst in this period related to data mining and classification, NLP and computational linguistics. In the same period, across all countries evidence showed bursting of skills related to machine vision, including image recognition and processing, pattern recognition as well as motion planning.
• The burst behaviours observed in Canada and the US, and to a lesser extent in the UK and Singapore, further confirm a growing trend in the demand for deep learning-related skills, most notably in 2017 and 2018.
• In terms of occupational groups under which AI-related jobs are advertised, many AI-related jobs belonged to the “Professionals” and “Technicians and associated professionals” categories.
• Moreover, with the exception of “Skilled agricultural, forestry and fishery workers”, which nevertheless are insufficiently represented in BGT data (see Cammeraat and Squicciarini (2021, forthcoming[2]) ), a growing number of AI- related jobs openings was posted online over time, in relation to all occupational groups.
• Skills related to AI appear to be in demand across almost all sectors of the economy, though to varying degrees. The group of sectors “Information and Communication” (J), “Financial and Insurance Activities” (K) and “Professional, Scientific and Technical Activities” (M) ranked at the top in terms of AI-job intensive sectors, in all countries considered.

Source: OECD iLibrary | Demand for AI skills in jobs: Evidence from online job postings