Careers & University / Big Questions / AI and work
From the research interviews we hold · updated 09 Nov 2026
The research points to a clear answer: technical knowledge alone won't be enough, because the skills that hold value are the ones AI struggles with — judging outputs, spotting errors, and knowing a field deeply enough to catch when AI is wrong. At the same time, several findings warn that AI's spread is uneven and uncertain, so the safe bet is combining deep subject knowledge with the ability to work alongside these tools.
Word for word, from interviews with the researchers. Each opens the moment they say it.
As AI takes over technical work, differentiating oneself through human factors becomes increasingly important for employment.Mark Shapiro, University of Miami — Professional Credibility, Early-Career Performance, and Workplace Reputation in Business · 10:54Paper
The key challenge for workers in 2026 is understanding both the capabilities and limitations of AI, and prior expertise enables one to evaluate AI-generated answers by recognizing when a small portion of the response is inaccurate.Mark Shapiro, University of Miami — Professional Credibility, Early-Career Performance, and Workplace Reputation in Business · 12:04Paper
Domain knowledge is essential for evaluating the validity, efficacy, and accuracy of AI outputs.Garry Falloon, Federation University Australia — Human-GenAI knowledge building: zones of cognitive partnership · 3:55Paper
Domain experts with about ten years of experience are influenced by AI decisions to the same degree as novices.Leonard Boussioux, University of Washington — The Narrative AI Advantage? · 15:24
Friction in these models arises from not knowing what these technologies are leading us to and what kind of skills will be needed.Sir Christopher Pissarides, London School of Economics — Markets with Search Friction · 6:32Paper
Programmers are needed to break big stories, and the new skill set for the future is programming, including machine learning.Janet Jones, London South Bank University — Technological changes affecting journalism · 8:03
The question assumes there is one clear answer, but the research splits into two camps. One camp says yes — certain human skills will keep mattering, and they are specifically the skills AI is bad at. The other camp says the ground is shifting so fast that nobody can confidently name the winners yet.
Start with the 'yes' side. Mark Shapiro [2] argues that as AI absorbs technical work, standing out through human factors becomes more important for getting hired. He builds on this in [4]: the key skill for workers is understanding both what AI can do and where it fails, and having enough prior expertise to notice when a small part of an AI answer is wrong. Garry Falloon [28] makes the same point from an education angle — you need real domain knowledge to judge whether an AI output is valid, effective, and accurate. These three line up: deep subject knowledge plus critical evaluation is the durable combination.
Now the pushback. Leonard Boussioux [5] found that domain experts with about ten years of experience were swayed by AI decisions just as much as novices. That undercuts the comforting idea that expertise automatically protects you — if experts are just as influenced, then 'know your field well' may not be the shield people assume. Sir Christopher Pissarides [25] adds a structural warning: the friction in these models comes from not knowing where the technologies are leading or what skills will be needed. In other words, the question itself may be unanswerable right now.
Janet Jones [12] sits in between. She says programmers are needed to break big stories and that programming, including machine learning, is the new skill set — which supports the 'yes' answer but points to a technical skill rather than a human-judgment one. That is a different bet from Shapiro's and Falloon's.
Why do researchers disagree? They are looking at different things. Shapiro and Falloon study how people evaluate AI outputs in workplace and classroom settings. Boussioux measures influence in decision experiments, which captures something subtler than skill. Pissarides looks at labour-market friction across whole economies. Different countries, different measures, different time horizons — a lab finding about experts being swayed is not the same as a hiring trend. The honest reading: judgment and domain depth look valuable, but don't assume expertise alone will protect you, and expect the specific skill list to keep moving.