From Reacting to Anticipating: The AI Skills Gap

Defining the Problem: AI is changing work

In recent years, we have seen iconic figures in the AI market predict that within a couple of years, some types of jobs will disappear. Last year, Anthropic’s CEO, Dario Amodei, stated that in the next five years entry-level posts in industries such as legal, consultancy, and finance “will be completely wiped out”.¹ Earlier, in 2023, Elon Musk, CEO of Tesla and SpaceX, said that “there will come a point where no job is needed … AI will be able to do everything.”²

A survey conducted by the Policy Institute and the Institute for Artificial Intelligence at King’s College London, released in May 2026, shows that about 56% of the public agree with Amodei’s prediction. It also shows that 75% of the public think AI will lead to widespread unemployment.³⁻⁴

At the same time, technology also plays a role in creating new opportunities. The reality appears more complex than the sole narrative of “machine taking jobs”, with continuous job destruction and creation driven by a myriad of different interacting elements.⁵ The complexity becomes more prominent when one considers that AI can affect employment through multiple channels, not only fully automating entire work pipelines, but also augmenting, reconfiguring tasks, and complementing or substituting on-demand skills.

Along these lines, the International Labour Organisation (ILO) concluded in its 2025 update on GenAI and jobs that even though, across the world, one in four workers are in occupations with some degree of exposure, most jobs tend to be transformed rather than made redundant.⁶ Other separate studies, with different methodologies and datasets, converge on similar findings, i.e., where job changes in the face of AI are real and evolving but remain uneven and are not yet indicative of a labour-market collapse.⁷

Therefore, reflecting on the future requires us to move beyond the question of whether jobs will disappear altogether or whether society will have enough jobs, and also look into which tasks are changing, and whether workers will have the skills needed for the jobs that emerge, and those that remain but are transformed. Such an understanding is relevant so governments can prepare for the new demands of employers, adopting a proactive policy approach that can better support the economic transition.

In this context, the present article tackles the topic: “How Global AI Policy Is Rewriting the Rules of Work and How to Stay Ahead of the Curve”, which was recently a theme of debate at the 2026 Global Work Conference.⁸

The Two Dimensions of the Skills Question

When considering that the adoption of AI is transforming job tasks and organisational requirements, two dimensions need to be considered. Despite being related to the same phenomenon of shifting towards an AI-powered economy, these facets are distinct from one another and call for different institutional responses.

The first dimension concerns challenges related to the larger population of workers, who will need to have the capabilities to adapt to changing occupations, expectations and responsibilities, while working alongside AI. Having the skills to build AI and the skills to work in an economy increasingly shaped by it are two different parts of the same puzzle, referred to as the “skills gap”. Both can get in the way of organisations being able to adopt AI effectively, but they do so in slightly different ways and should not be conflated.

The second dimension concerns challenges related to the supply of specialised technical expertise to develop, implement, and govern AI – the so-called “AI workforce” (i.e., workers with statistics, computer science, machine-learning and AI Governance skills).⁹

The Organisation for Economic Co-operation and Development (OECD) suggests that advanced technical AI skills will be required for less than 1% of workers.¹⁰ Similarly, researchers at the ILO argue that the biggest concern should be less in investing only in AI expertise and more in enabling the broader workforce to work alongside AI.¹¹ While understanding the risks that AI poses to their work and knowing how to use it properly within their employment context are important AI literacy skills for the overall workforce, academic and policy evidence points towards the need for a broader skill-set, including digital and data skills, critical thinking, creativity, domain expertise, and the ability to adapt as tasks change.¹²

Not every worker will be part of the specialised “AI-workforce”, but that does not mean AI will not have a role in their jobs. As the World Economic Forum (2025) points out,¹³ employers continue to value skills such as analytical thinking, resilience, flexibility and agility, leadership, creative thinking and technological literacy, alongside AI and big data skills.¹⁴

The implication is that bridging the skills gap is not about enabling more people to build AI and to know how to code. It is also about preventing the rest of the workforce from becoming technologically dependent without developing the ability to understand, use, and critically evaluate AI.¹⁵

How Governments are Responding to the Issue

The OECD has recently published a review of policy measures related to AI in the labour market, covering G7 countries, the European Union (EU), and selected Latin American countries.¹⁶ The report shows that, alongside general unemployment assistance for those who have already lost work to automation, several surveyed countries offer targeted support to workers most exposed to automation risk, and skills programmes addressing the “skills gap” identified above.

However, most policy initiatives focus on developing specialised AI skills rather than general AI literacy. Together with the fact that the report is silent on other soft skills, it can indicate that most policymakers are tackling the issue through only one of its dimensions. This shows that the two mapped dimensions of the problem are hard to engage with through a single “AI workforce policy”, and it may require a deeper (and larger) discussion about education systems, including teaching methods and curriculum.

Responses and their Institutional Context: UK and EU

When separately analysing the United Kingdom (UK) and EU as jurisdictions where broader AI literacy appears as a specified concern, one can see that the way governments respond to the issue is influenced by their institutional context, including broader strategies encompassing economic and social goals – which modify how the problem itself is framed.

In the UK, where the government adopts a principles-based approach to AI with sectoral implementation,¹⁷ the AI Opportunities Action Plan (2025) states that the goal is building “a deeper pool of AI skills and talent” to support cross-economy AI adoption, as a pillar of economic growth.¹⁸ While framing the issue towards AI capability across the economy generally, the UK identifies literacy as a key point for employability and inclusion.¹⁹

In the EU, the strategy focuses on AI-specific legislation to safeguard human rights. Literacy appears as an enforceable duty. The EU AI Act required organisations to ensure a sufficient level of AI literacy among staff interfacing with AI.²⁰ In July 2026, that duty was softened to only “support the development of AI literacy”, with no specific level guaranteed.²¹ The rule was adopted, then weakened, signalling to the market that literacy is a valued capability, but showing uncertainty about its implementation, while moving obligations from organisations to European and national levels.

While the UK illustrates that the “skills gap” problem can be framed through broader policy goals, bringing about questions related to government capacity to deliver it and keep track of the changing landscape, the EU shows that regulations can also impact the framing.

This means that any AI-related policy is layered unevenly onto pre-existing frameworks, where the future of work gets addressed differently depending on which policy philosophy a government adopts.

Closing Reflections: from reacting to anticipating

Given this uneven response, the question concerns its practical implications. There is uncertainty about future employment distribution, but skills, tasks and organisational requirements are already changing, as policy attempts to catch up and shape that transition. Recognising the interactions between these factors is key to keeping up with the curve, which is transformed by technological developments, but not alone. Therefore, it is for us to ask: What capabilities will become valuable as AI, organisations and the rules governing them evolve together?

As seen, governments’ main focus currently lies in developing specialised AI skills, but employers needs go beyond that by combining AI literacy and soft skills. For governments, this indicates an urgency in broadening the discussion and rolling out broadly accessible programs complementing organisations’ efforts in a broad upskilling. For workers, evidence indicates that value is concentrating at the intersection of domain expertise, AI fluency, and the ability to read the rules forming around both.

When it comes to AI literacy more specifically, a few jurisdictions are already addressing the concern through different strategies. However, what “meaningful” literacy looks like in practice, and how to deliver it, is undefined, and waiting until a standard comes along risks leaving organisations and individuals behind, indicating that a proactive approach can be more suitable.

There is no “one size fits all” solution, but shifting from reaction to anticipation is what the evidence supports. As discussed above, reducing AI upskilling to either basic AI literacy or in-depth technical knowledge risks being short-sighted. Societies – and, by extension, economies – that seek to harness AI for economic growth and the betterment of their members need to adopt a much broader approach to AI skills development.

References

¹ VandeHei, J. & Allen, M. (2025) ‘Behind the Curtain: A white-collar bloodbath’, Axios, 28 May. Available at: https://www.axios.com/2025/05/28/ai-jobs-white-collar-unemployment-anthropic

² CNBC (2023) ‘Tesla boss Elon Musk says AI will create situation where no job is needed’, 2 November. Available at: https://www.cnbc.com/2023/11/02/tesla-boss-elon-musk-says-ai-will-create-situation-where-no-job-is-needed.html

³ King’s College London (KCL) (2026) ‘One in five Britons think AI will create civil unrest, study finds’. Available at: https://www.kcl.ac.uk/news/one-in-five-britons-think-ai-will-create-civil-unrest-study-finds

⁴ See also: Advisory, Conciliation and Arbitration Service (Acas) (2025) ‘1 in 4 workers worry that AI will lead to job losses’, 28 April. Available at: https://www.acas.org.uk/1-in-4-workers-worry-that-ai-will-lead-to-job-losses; Security Brief UK (2026) ‘AI fears push one in three workers to quit industry’. Available at: https://securitybrief.co.uk/story/ai-fears-push-one-in-three-workers-to-quit-industry; McClain, C. & Park, E. (2026) ‘Young adults in the U.S. are increasingly wary of AI, concerned it will take jobs’, Pew Research Center, 18 August. Available at: https://www.pewresearch.org/?p=358582

⁵ De, D. (2026) ‘Machines – job destroyers or liberators?’, The Monday Briefing, Deloitte UK, 23 February. Available at: https://www.deloitte.com/uk/en/blogs/monday-briefing/machines-job-destroyers-or-liberators.html

⁶ Gmyrek, P., Berg, J., Kamiński, K., Konopczyński, F., Ładna, A., Nafradi, B., Rosłaniec, K. & Troszyński, M. (2025) Generative AI and Jobs: A 2025 Update. International Labour Organization (ILO), 20 May. Available at: https://www.ilo.org/publications/generative-ai-and-jobs-2025-update

⁷ Acemoglu, D., Autor, D., Hazell, J. & Restrepo, P., ‘Artificial Intelligence and Jobs: Evidence from Online Vacancies’, Journal of Labor Economics, Vol. 40, No. S1, pp. S293–S340, April 2022. https://www.journals.uchicago.edu/doi/10.1086/718327; Humlum, A. & Vestergaard, E., ‘Still Waters, Rapid Currents: Early Labor Market Transformation under Generative AI’, NBER Working Paper 33777, 2025. https://www.nber.org/papers/w33777; Hampole, M., Papanikolaou, D., Schmidt, L. & Seegmiller, B., ‘Artificial Intelligence and the Labor Market’, NBER Working Paper 33509, February 2025. https://www.nber.org/papers/w33509

⁸ Global Work Conference. Available at: https://globalworkconference.com/

⁹ Green, A. & Lamby, L. (2025) ‘The Characteristics of the Artificial Intelligence Workforce across OECD Countries’, The Indian Journal of Labour Economics, Vol. 68, No. 2, pp. 541–568. Springer / Indian Society of Labour Economics (ISLE), June. DOI: 10.1007/s41027-024-00549-7.

¹⁰ OECD (2026), AI and skills: What we know so far, OECD Publishing, Paris. https://doi.org/10.1787/f843b352-en

¹¹ Delaporte, I. & Liepmann, H. (2026) ‘Old Skills for new technologies?’, International Labour Organization (ILO), 14 July. Available at: https://www.ilo.org/resource/article/old-skills-new-technologies

¹² Babashahi, L. et al. (2024), ‘AI in the Workplace: A Systematic Review of Skill Transformation in the Industry’, Administrative Sciences, Vol. 14, No. 6, Article 127, 16 June. Available at: https://www.mdpi.com/2076-3387/14/6/127

¹³ World Economic Forum (2025), Future of Jobs Report 2025, 7 January. Available at: https://www.weforum.org/publications/the-future-of-jobs-report-2025/

¹⁴ See also: Bardak, U., Pouliakas, K., Hippe, R., Sostero, M., Tasiovasilis, M., Kizu, T., Tumurchudur-Klok, B., Strietska-Ilina, O., Martin Lataix, J.I., Iriarte, R. & Zhang, W. (2026) Changing Landscape of Skills in the Age of AI. International Labour Organization (ILO), 13 August. doi: 10.2816/7083675. Available at: https://www.ilo.org/publications/changing-landscape-skills-age-ai

¹⁵ See: Bellamy, J., Douglas, J., Wickett-Whyte, J. & Bransden, N. (2026) AI Skills for Life and Work: Summary report. Ipsos, supported by the Department for Science, Innovation and Technology (DSIT) and the Department for Culture, Media and Sport (DCMS), 28 January. Available at: https://www.gov.uk/government/publications/ai-skills-for-life-and-work-summary-report

¹⁶ OECD (2026), Recent policy developments on AI in the labour market, OECD Artificial Intelligence Papers, No. 63, OECD Publishing, Paris. https://doi.org/10.1787/c0ffced7-en

¹⁷ Department for Science, Innovation and Technology (DSIT) (2024) Implementing the UK’s AI Regulatory Principles: Initial Guidance for Regulators. GOV.UK, 6 February. Available at: https://assets.publishing.service.gov.uk/media/65c0b6bd63a23d0013c821a0/implementing_the_uk_ai_regulatory_principles_guidance_for_regulators.pdf

¹⁸ Department for Science, Innovation and Technology (2025) AI Opportunities Action Plan. GOV.UK, 13 January. Available at: https://www.gov.uk/government/publications/ai-opportunities-action-plan/ai-opportunities-action-plan

¹⁹ Bellamy, J., Douglas, J., Wickett-Whyte, J. & Bransden, N. (2026) AI Skills for Life and Work: Summary report. Ipsos, supported by the Department for Science, Innovation and Technology (DSIT) and the Department for Culture, Media and Sport (DCMS), 28 January. Available at: https://www.gov.uk/government/publications/ai-skills-for-life-and-work-summary-report

²⁰ European Union (2024) Regulation (EU) 2024/1689 of the European Parliament and of the Council, Article 4 (Artificial Intelligence Act).

²¹ European Union (2026) Regulation (EU) 2026/1744.