AI Skill Advisor Helps Freelancers Reskill, but Rankings Can Backfire

Image: "Laptop and small plant on a desk in a modern workspace during daylight" by Shixart1985 (CC BY 2.0) via Wikimedia Commons, recoloured.
Which skills should a freelancer learn next as AI reshapes labor markets? A 2026 Information Systems Research study builds an AI-enabled skill advisor, FESA, that recommends skills tied to the tasks freelancers want. It outperformed existing approaches on real platform data, and an experiment showed that revealing low market rankings can reduce motivation to reskill.
Designing AI that tells workers what to learn
Online labor platforms are very good at matching: employers find workers and workers find jobs. What they rarely offer is advice on how a worker should build skills to stay competitive. As AI shifts the value of different skills, that gap matters more. According to Liu, Yu, He, Wen and Chai (Information Systems Research, 2026), the right starting point is the freelancer's own goals rather than the platform's matching needs.
The team developed a freelancer-centric skill advisor, FESA, that recommends skills based on the tasks a freelancer wants to win. It supports two reskilling strategies: exploration, which means picking up new skills, and exploitation, which means deepening expertise a freelancer already has. The design question is not only which skills to suggest, but which strategy a given recommendation serves.
The evaluation runs in two parts. Using real-world data from a major online labor market, the authors report that FESA substantially outperforms existing approaches in identifying skills that improve freelancers' competitiveness. A controlled experiment then examined how people respond to the advice: participants were more motivated to pursue FESA's personalized recommendations than generic recommendations based on popular skills.
The most interesting result is a design lesson. The experiment shows that more information is not always better. When the system revealed a freelancer's low market ranking, motivation to reskill fell, particularly under exploration strategies, where the worker is asked to move into unfamiliar territory. Transparency that seems helpful in principle can discourage the very behavior the tool is meant to encourage.
The published abstract reports results from one major labor platform and a controlled experiment, so readers should check the full paper for sample details, effect sizes and how far the findings generalize to other platforms, occupations and longer time horizons. Motivation measured in an experiment is also not the same as observed reskilling and later earnings.
Why it matters: For information systems and management researchers, this is a clean example of design science that pairs an algorithmic contribution with a behavioral test, and the behavioral test changes the design advice. For anyone studying AI-driven recommendations, from career tools to learning platforms, it is a reminder to measure how feedback framing affects motivation, not only how accurate the recommendation is.
Liu, H., Yu, S., He, S. X., Wen, W., & Chai, Y. (2026). Improving Competitiveness: A Freelancer-Centric Approach for Recommending Reskilling Strategies. *Information Systems Research*. Read the original. Independent editorial summary of third-party research.





Comments