Reskilling the Workforce for an AI-Driven Economy

The Urgency of Workforce Reinvention

"AI is going to reshape every industry and every job," says Reid Hoffman, Co-founder of LinkedIn.¹ The pace of change in AI in the technology industry is simply unprecedented. Rapid research and development in industry and academia are driving this, which has inevitably started to percolate to other sectors and domains. In this environment, there is a certain apprehension about AI leading to large-scale job losses. While this is more apparent in the technology industry, it is not without reason that workers in other sectors are also asking what it means to be skilled and prepared for the AI era. This chapter will explore how AI is impacting various domains and industry verticals. It will examine current reskilling efforts, both success stories and reasons for failures, and make policy recommendations to prepare the global workforce for the AI era.

Technical Innovation and the Drivers of Change

The current pace of change related to AI in the technology industry and in terms of the research coming from academia is leading to phenomenal changes and is reshaping companies. I witnessed the boom of cloud computing and distributed systems in the last decade, but that does not hold a candle to the current pace of change and rapid evolution in AI.

One area of rapid innovation is the increasingly specialized capabilities of large language models (LLMs). AI labs², such as OpenAI, Anthropic, Google, Meta, etc., are racing against each other to post-train their LLMs in highly specialized domains. Examples include languages and dialects such as Eastern Levantine Arabic and North Vietnamese. Other examples that go beyond natural language processing include musicology and radiological images for lung cancer and pneumonia.

Machine learning techniques such as Reinforcement Learning with Human Feedback, Direct Preference Optimization, etc., drive an entire industry worth billions of dollars. Companies such as Scale AI, Surge AI, Turing, Mercor, etc., provide ground truth labeled data and expert labelers who can help generate new datasets for post-training of LLMs. These expert labelers and the datasets they are helping provide are in very diverse fields, such as K-12 education, inorganic chemistry, linguistics, coding, etc. Indeed, in several fields, AI labs have successfully post-trained LLMs to PhD and even professor-level expertise.

New Skills for a New Economy

The rapid innovation in AI has led to machine learning models, specifically LLMs, becoming highly adept in various fields. This can lead to apprehension about job replacements. Still, it is more of a reason for optimism and excitement about the positive changes AI can bring through new economic opportunities.

Consider some inspiring examples across different domains where AI has been successfully applied.

Health Care

The FDA manages a list of cleared AI/ML-enabled medical devices, signifying AI's growing real-world application and regulatory acceptance in clinical diagnostics and patient care.³ In medical imaging for diabetic retinopathy, AI analyzes retinal fundus photographs for detection with accuracy comparable to human ophthalmologists, enabling earlier screening and treatment.⁴ In drug discovery, Insilico Medicine used AI to help identify a novel therapeutic target and generate a new drug candidate for Idiopathic Pulmonary Fibrosis, which has now progressed to human clinical trials, drastically shortening typical discovery timelines.⁵ AI also contributes to enhanced 3D visualization, improved instrument control, and data analytics in robotic surgical systems, aiming for more precise and less invasive procedures. The Da Vinci surgical system is a good example of using AI for this specific use case.⁶

Finance

Mastercard uses AI to analyze real-time transaction data, improve fraud detection, and reduce false positives for legitimate purchases.⁷ For many RoboAdvisors, such as Betterment, etc., AI algorithms enable the automation of financial planning, portfolio construction, asset rebalancing, and tax-loss harvesting, lowering the cost of personalized investment management.

Manufacturing and Agriculture

The manufacturing industry has embraced predictive maintenance as part of various industrial processes. Companies like C3 AI, Siemens, and Aveva have driven global adoption of various predictive maintenance products in several industry verticals. Landing AI's platform, which uses AI-driven computer vision to automate visual inspection processes in manufacturing, identifies defects with greater speed and consistency than manual methods.⁸ In the agriculture sector, John Deere's See & Spray technology uses computer vision and AI to differentiate between crops and weeds in real-time. This enables precise herbicide application only where needed, reducing chemical usage and costs.⁹ Farmers and gardeners also benefit from adopting AI to detect plant diseases. An example is the Plantix app, which uses AI-powered image recognition to analyze photos of crops taken by farmers, helping to quickly diagnose plant diseases, pest infestations, and nutrient deficiencies.¹⁰

Media and Entertainment

The adoption of AI in the media and entertainment sector, specifically in user personalization, is one of the more apparent effects of the AI revolution. Meta employs extensive AI and machine learning algorithms to analyze user data, including specific interests and behaviors. This is utilized to deliver targeted content and advertisements across its platforms.¹¹ Netflix uses sophisticated AI algorithms to analyze viewing history and preferences, provide personalized content recommendations, and even tailor artwork to increase user engagement.¹² The easy access to multimodal LLMs has ushered in a world of creativity and enhanced content creation, including ad creation.

Science and Research

DeepMind's AlphaFold AI system predicts the 3D structure of proteins from their amino acid sequence with remarkable accuracy, significantly accelerating biological research and drug discovery.¹³ In climate modeling, NVIDIA's Earth-2 platform leverages AI and supercomputing to create a high-resolution digital twin of Earth, enabling improved climate and weather simulations for better prediction and mitigation strategies.^{14 15} These are just a few examples of AI enabling scientists and researchers to make rapid advancements.

Education, Autonomous Vehicles, and Transportation Logistics

AI's impact can also be observed in other domains, such as the very important sector of education. Khan Academy's AI-powered teaching assistant and tutor, Khanmigo, offers personalized guidance, workflow simplification, and guidance for critical problem-solving while assisting educators with lesson planning and progress tracking.¹⁶ In self-driving, Waymo's AI enables self-driving vehicles to perceive their surroundings, predict the behavior of other road users, and navigate complex urban environments autonomously. In transportation logistics, for instance, UPS's On-Road Integrated Optimization and Navigation (ORION) system uses AI to determine the most efficient delivery routes, reducing miles driven, fuel consumption, and emissions.¹⁷

In all of these examples across domains and industry verticals, AI enables greater efficiency, productivity, and output from human endeavor. And therein lies the alleviation of any apprehension about AI. Despite all the advances and the pace of innovation, AI is not there yet to truly replicate unique human skills and, in fact, the lived human experience. Our sentience makes us human, more than our sense of reasoning and perception. Despite all the talk about Artificial General Intelligence (AGI), we are not at a point where we can expect AI algorithms that go beyond token prediction to achieve sentience. This sentience, along with the accompanying human traits of empathy, friendship, happiness, and sadness, among others, drives our most significant endeavors. The unpredictability element also makes the lived human experience unique and eventful in our history. Human beings certainly possess many skills that AI currently lacks. These include complex critical thinking, ethical judgment and moral reasoning, nuanced communication, and persuasion. These unique human skills enable us to ensure that leveraging AI in conjunction with them can lead to new economic opportunities.

To a large extent, LLMs need guidance (or prompting) from human beings to discover or create anything new. In the creative field, human skills of imagination, storytelling, artistic vision, and understanding cultural nuances, combined with generative AI for different modalities, can act as an assistant, rapidly producing drafts, variations, or components. For complex problem solving, human skills of defining ambiguous problems, critical thinking, strategic foresight, and contextual understanding can leverage AI's ability to analyze massive datasets to identify correlations, anomalies, and potential solutions that humans might miss. These AI human skills can be further augmented when used with reasoning models. Human skills of ethical reasoning, moral judgment, accountability, and understanding societal impact can lead to the development of AI tools that can help audit algorithms for bias or monitor AI systems for unintended consequences. These explorations barely scratch the surface of the economic opportunities that AI can lead to.

Why Reskilling Efforts Succeed and Why They Fall Short

There are several reskilling efforts underway from both the public and private sectors.

Accenture, a global professional services company, has been heavily investing in upskilling its global workforce in "New IT," with a significant component being Artificial Intelligence, including Generative AI. This upskilling initiative ensures that Accenture consultants can advise clients and implement AI solutions effectively. ¹⁸In late 2023, Amazon pledged its "AI Ready" initiative, aiming to provide free AI skills training to 2 million people globally by 2025. This expands on their "Upskilling 2025" commitment ($1.2 billion investment to upskill 300,000 employees). "AI Ready" includes new and existing free AI and generative AI courses on AWS Skill Builder, a new AWS Generative AI Scholarship program, and collaborations with organizations like Code.org. The training ranges from foundational AI concepts for business leaders to advanced skills for developers.¹⁹ Microsoft has a multi-pronged approach. They offer extensive AI training through Microsoft Learn, and LinkedIn (owned by Microsoft) launched a "Career Essentials in Generative AI" professional certificate and numerous AI courses. Microsoft also partners with public sector organizations to deliver AI skills training.²⁰

In the public sector, a key example is that of Singapore, who have a comprehensive national lifelong learning movement, SkillsFuture. Within this, AI Singapore (AISG)²¹, a national AI program, runs several key initiatives targeting different labor segments. The UK's National AI Strategy²²includes a strong focus on skills, including the specific business guidance developed by the Alan Turing Institute and others.²³

Programs addressing current and anticipated AI skill shortages in the job market tend to succeed. Learners are more motivated and likely to complete training when they see a clear path to better job opportunities or career advancement. This has been demonstrated by programs like Amazon's upskilling and Singapore's AIAP, which have high placement rates. Industry-recognized certifications such as those from cloud providers like GCP, AWS, and Azure prove skills that enhance training value. Public and private sector collaborations ensure curriculum relevance, access to expertise, funding, and employment pathways. Government backing (like in Singapore and the UK) provides strategic direction, resources, and credibility. The AI field is evolving very rapidly. Successful programs regularly update their content to reflect the latest tools, techniques, and industry needs - especially in areas like generative AI.

Training programs that teach outdated skills or do not align with what employers are actively seeking will lead to poor employment outcomes and disillusioned learners. Generic AI courses without specialization, a clear link to job roles, or hands-on experience with AI tools and real-world datasets can be less effective. Even programs sponsored by companies for their employees can fail if management doesn't actively support employees' learning time, create opportunities to apply new AI skills, or recognize their new competencies with new roles or responsibilities. Initial reskilling programs can lack the necessary funding, resources, or sustained support to expand effectively or achieve broader success. Establishing frameworks for ongoing education, measuring success metrics, and evolving programs based on needs is crucial.

An article in Nature analyzed several research papers on ChatGPT's effect on student learning.²⁴ This study offers important lessons for reskilling efforts. Problem-based learning models and tailoring content of skills programs for STEM applications and communications are essential for course design. Using AI as an intelligent tutor and combining it with human mentorship leads to more positive outcomes.

Policy Recommendations: Building Human-Centered Reskilling Infrastructures

The following are some recommendations for building human-centered reskilling programs.

• Affordable and Flexible Learning Pathways: Provide financial aid and economic incentives for AI education programs, particularly for unrepresented communities. Promote modular learning programs with skills that can be added on, with the flexibility of online and part-time access. Support non-traditional learners and rural populations, especially in emerging markets.
• Easy Access to AI Literacy: Promote AI literacy starting at the K-12 level and include it in adult education programs. Provide free online resources for self-paced learning.
• Personalized Learning and Career Guidance: Tailor AI reskilling programs for specific skills and experience levels while taking into account any existing familiarity with AI tools. For added impact, combine this with human mentorship and career counselors.
• Practical Training with Industry Partnership: Establish frameworks for co-designing and accrediting reskilling programs with strong industry involvement to ensure curricula are relevant, hands-on, and lead to in-demand skills. Promote apprenticeships and work-integrated learning, increasing the likelihood of employment post-training and providing practical and current skills.
• Support for Continuous Learning: Launch public awareness campaigns to normalize and encourage continuous learning and adaptation throughout one's career. Recognize and reward individuals and organizations that champion lifelong learning.
• Incentives for Reskilling: Provide incentives such as tax breaks and employee bonuses for employers that offer reskilling programs, and for employees who enroll in them. Provide wrap-around services like mentorship, childcare, job placement support, etc.
• Labor Market Intelligence: Invest in substantial data collection and analysis systems to monitor changing skill demands related to AI, identify new job roles, and predict future needs. Make this information publicly accessible. Doing so will enable policymakers, educators, and individuals to make informed decisions about investments in reskilling.

Mechanisms for Scale and Accountability

Mechanisms to achieve scale and accountability in AI reskilling programs can ensure they are effective, equitable, and responsive to evolving needs. This involves robust governance models, clear metrics for success, and supportive policies.

Mechanisms for scale include public-private partnerships. This involves state entities closely collaborating with tech companies to provide platforms and content, educational institutions to deliver training, and industry associations to define skill needs and offer job placements. Training delivery should leverage massive open online course (MOOC) platforms. This allows for the simultaneous delivery of standardized foundational content to many learners at a relatively low marginal cost. Following the train-the-trainer model, cohorts of educators, community leaders, and corporate trainers should be upskilled and trained to deliver AI reskilling programs locally or within their organizations. Organizations should also look to develop nationally or internationally recognized foundational AI curricula and stackable

micro-credentials. This simplifies program development, ensures a baseline quality, and allows learners to accumulate skills incrementally from various providers. There should also be efforts to integrate AI reskilling modules into existing vocational training centers, community colleges, public libraries, and adult education programs.

Governance Models for Accountability should include multi-stakeholder bodies, independent oversight and auditing, objective performance management, objective measurement systems, transparent reporting, and public relations. National or regional AI skills councils should be formed from representatives of government, industry, academia, worker unions, and civil society to address the various challenges and opportunities of artificial intelligence. These councils will set the strategic direction, define quality standards, manage funding allocation, monitor program effectiveness, and ensure that AI programs meet diverse societal needs. Mandate regular public reporting on program enrollments, completion rates, demographics of learners, employment outcomes, and expenditure to ensure accountability and create avenues for course correction when necessary.

Metrics for measuring success may include learner-centric metrics such as skills acquisition, completion rates, and NPS scores. Employment and economic metrics can include job placement rates, employer and industry metrics, and program efficiency metrics. Employer satisfaction and skills gap reduction data can measure industry metrics, while cost per successful outcome and time to gain employment could be metrics for program efficiency.

Policies are critical in encouraging reskilling initiatives and providing the right incentives. Financial incentives should be provided for both individuals and employers. Training providers should be given grants and training opportunities. Regulatory frameworks need to be established for the administration and recognition of qualifications. Public awareness and advocacy campaigns are also essential. Data gathering and analysis are crucial for developing labor market intelligence. It is important to promote ethical and responsible AI reskilling. Policies should also be adapted depending on the target market, specifically in developed vs. emerging economies.

Reimagining Work as Human Flourishing in the AI Era

“The first thing I would do is to learn AI,” said Jansen Huang, the CEO of Nvidia, if he were to be a student again.²⁵ Reskilling initiatives must start by integrating ethical AI use into elementary or middle school curricula. Then, as we have discussed, public-private initiatives at the industry or state level with clear incentives, measurable goals, and governance models can significantly impact ensuring that the workforce of the future is set up for success in the AI era. We have been through several transformations in our history, whether economic, political, or technological. Looking back, it would be safe to say that we are better off now than we were several hundred years ago or even a hundred years ago. We will adapt, evolve, and find new opportunities to be even more successful going into the future. AI will play an extraordinary role in enabling us to do so.

Endnotes:

¹ Harroch, Dominique A, and Richard Harroch. “15 Quotes on the Future of AI.” TIME. redesign, April 25, 2025. https://time.com/partner-article/7279245/15-quotes-on-the-future-of-ai/.

² Companies developing LLMs

³ Center. “Artificial Intelligence and Machine Learning (AI/ML)-Enabled Medical D.” U.S. Food and Drug Administration, 2025.

https://www.fda.gov/medical-devices/software-medical-device-samd/artificial-intelligence-and-machine-learning-ai ml-enabled-medical-devices.

⁴ Varun Gulshan, Lily Peng, Marc Coram, Martin C Stumpe, Derek Wu, Arunachalam Narayanaswamy, Subhashini Venugopalan, et al. “Development and Validation of a Deep Learning Algorithm for Detection of Diabetic Retinopathy in Retinal Fundus Photographs.” JAMA 316, no. 22 (November 29, 2016): 2402–2. https://doi.org/10.1001/jama.2016.17216.

⁵Insilico.com. “First Generative AI Drug Begins Phase II Trials with Patients | Insilico Medicine,” 2023. https://insilico.com/blog/first_phase2.

⁶ Open Medscience. “The Da Vinci Technology: Pioneering a New Era in Medical Imaging and Patient Care.” Open MedScience, April 28, 2023. https://openmedscience.com/the-da-vinci-technology-pioneering-a-new-era-in-medical-imaging-and-patient-care/.

⁷ Mastercard. “Mastercard Accelerates Card Fraud Detection with Generative-AI Technology.” Mastercard.com, November 2, 2024. https://www.mastercard.com/news/press/2024/may/mastercard-accelerates-card-fraud-detection-with-generative-ai-t echnology/.

⁸ LandingAI. “LandingLens,” May 7, 2025. https://landing.ai/landinglens.

⁹ Deere, John. “See & SprayTM Ultimate | Precision Ag | John Deere US.” Deere.com, 2018. https://www.deere.com/en/sprayers/see-spray-ultimate/.

¹⁰ PEAT GmbH. “Plantix.” Plantix. PEAT GmbH, 2025. https://plantix.net/en/.

¹¹ Meta.com. “The AI behind Unconnected Content Recommendations on Facebook and Instagram,” 2022. https://ai.meta.com/blog/ai-unconnected-content-recommendations-facebook-instagram/.

¹² Netflix Technology Blog. “Artwork Personalization at Netflix - Netflix TechBlog.” Medium. Netflix TechBlog, December 7, 2017. https://netflixtechblog.com/artwork-personalization-c589f074ad76.

¹³ Jumper, John, Richard Evans, Alexander Pritzel, Tim Green, Michael Figurnov, Olaf Ronneberger, Kathryn Tunyasuvunakool, et al. “Highly Accurate Protein Structure Prediction with AlphaFold.” Nature 596, no. 7873 (July 15, 2021): 583–89. https://doi.org/10.1038/s41586-021-03819-2.

¹⁴ Caulfield, Brian. “GTC Wrap-Up: NVIDIA CEO Outlines Vision for Accelerated Computing, Data Center Architecture, AI, Robotics, Omniverse Avatars and Digital Twins in Keynote.” NVIDIA Blog, November 9, 2021. https://blogs.nvidia.com/blog/nvidia-ceo-accelerated-computing-ai-omniverse-avatars-robots-gtc/.

¹⁵ NVIDIA. “NVIDIA Earth 2 Platform,” 2025. https://www.nvidia.com/en-us/high-performance-computing/earth-2/.

¹⁶ Khanmigo.ai. “Meet Khanmigo: Khan Academy’s AI-Powered Teaching Assistant & Tutor,” 2023. https://www.khanmigo.ai/.

¹⁷ Marr, Bernard. “The Brilliant Ways UPS Uses Artificial Intelligence, Machine Learning and Big Data.” Forbes, June 15, 2018. https://www.forbes.com/sites/bernardmarr/2018/06/15/the-brilliant-ways-ups-uses-artificial-intelligence-machine-lea rning-and-big-data/.

¹⁸ Accenture.com. “Accenture to Invest $3 Billion in AI to Accelerate Clients’ Reinvention,” 2023. https://newsroom.accenture.com/news/2023/accenture-to-invest-3-billion-in-ai-to-accelerate-clients-reinvention.

¹⁹ Amazon.com, Inc. “Amazon Announces ‘AI Ready,’ a New Initiative Designed to Provide Free AI Skills Training to 2 Million People by 2025.” Businesswire.com. Business Wire, November 20, 2023. https://www.businesswire.com/news/home/20231119572231/en/Amazon-Announces-AI-Ready-a-New-Initiative-De signed-to-Provide-Free-AI-Skills-Training-to-2-Million-People-by-2025.

²⁰ Microsoft Source. “AFL-CIO and Microsoft Announce New Tech-Labor Partnership on AI and the Future of the Workforce - Source.” Source, December 11, 2023. https://news.microsoft.com/source/2023/12/11/afl-cio-and-microsoft-announce-new-tech-labor-partnership-on-ai-an d-the-future-of-the-workforce/.

²¹ AI Singapore. “Home - AI Singapore,” May 3, 2022. https://aisingapore.org/.

²² for, Department. “National AI Strategy.” GOV.UK, September 22, 2021. https://www.gov.uk/government/publications/national-ai-strategy.

²³ for, Department. “New Business Guidance to Boost Skills and Unlock Benefits of AI.” GOV.UK, November 30, 2023. https://www.gov.uk/government/news/new-business-guidance-to-boost-skills-and-unlock-benefits-of-ai.

²⁴ Wang, Jin, and Wenxiang Fan. “The Effect of ChatGPT on Students’ Learning Performance, Learning Perception, and Higher-Order Thinking: Insights from a Meta-Analysis.” Humanities and Social Sciences Communications 12, no. 1 (May 6, 2025). https://doi.org/10.1057/s41599-025-04787-y.

²⁵ Jackson, Ashton. “Nvidia CEO: If I Were a Student Today, Here’s How I’d Use AI to Do My Job Better—It ‘Doesn’t Matter’ the Profession.” CNBC, May 17, 2025. https://www.cnbc.com/2025/05/17/jensen-huang-how-id-use-ai-to-do-my-job-better-if-i-were-a-student-today.html.