MIT SMR and BCG's AI panel majority calls human judgment the defining challenge of responsible AI
In MIT Sloan Management Review and BCG's fifth annual panel, 65% of AI experts agree human judgment is responsible AI's defining challenge. Its stated common ground is designing human-machine systems that keep trustworthy judgment available when it matters most.
The Board Room · Leadership desk
What happened
- A further 19% disagree or strongly disagree with the statement, and 15% neither agree nor disagree.
- The article cites new research suggesting AI overuse risks degrading cognitive and moral judgment, an effect it calls "epistemic de-skilling".
- It also points to New York City's first-in-the-nation ban on generative AI in schools, introduced after evidence-based concerns about children's cognitive and emotional development.
Why it matters
- decision Leaders have to decide which decisions get a person's judgment and which run at machine pace; the panel's majority view does not make that allocation for them.
- constraint Reviewer condition limits how much oversight is real: a reviewer facing the fatigue, bias or thin expertise Chow lists adds an approval step without adding judgment.
- exposure If de-skilling is real, the erosion falls on the cases Chow says need people most: novel situations, ethical trade-offs and context outside the model's training.
The 65% figure [2] tells a leader how many experts rank judgment as the central problem. It does not settle what to build. The article says the experts diverge on what human judgment means and how it differs from human oversight [4], so a new review step justified by "the panel agrees" rests on agreement about the diagnosis.
The sharpest skeptic sits inside the panel. Stefan Verhulst, GovLab's chief research and development officer, said, "Human judgment was never inherently trustworthy." [7] Yan Chow of Automation Anywhere said that "humans face fatigue, bias, lack of experience or expertise" [8]. The article's reply is that the answer is not simply to keep a human in the loop [6].
The cost side shows in Ben Dias's wording. Dias, IAG's chief AI scientist, said, "The defining challenge of responsible AI will most certainly be preserving trustworthy human judgment as increasingly capable AI systems operate at greater speed and scale." [10] In our view, judgment placed at every step gives up some of that speed and scale, and judgment placed nowhere leaves the de-skilling risk. This quarter's decision is which workflows fall in which group. Next quarter's consequence is whether reviewers who mostly approve AI output still have the skill to catch an error. If the de-skilling research holds [11], that skill is what erodes first, and we would treat the research as early because the article itself calls it new.
The evidence base is narrower than the headline number. The one policy the article cites, New York City's school ban, concerns children's development [12]. The text does not state how many panelists responded, and it breaks off before the practical recommendations the article says it offers [13].
What to watch
- The practical recommendations in the full MIT SMR article, which would show whether the panel favors review checkpoints or ongoing skill maintenance.
- Evidence on whether the de-skilling effect appears in adult employees, since the one policy example cited concerns schoolchildren.
- The reasons given by the 19% who disagree, which would show whether they reject the premise or only the definition of judgment.
Clarity's read
What the record supports and how the coverage leans. The claims behind it follow.
Reality
- Evidence35
- Adoption
- Insufficient
- Hype gap
- Insufficient
- Incentives
- Insufficient
- Confidence50
Claim ledger
Ranked by verification strength, evidence, and original report placement.
- [1]
For the fifth year in a row, MIT Sloan Management Review and Boston Consulting Group assembled an international panel of AI experts, academics and practitioners, on how responsible AI is being implemented across organizations.
- [2]
65% of panelists agree or strongly agree that the defining challenge of responsible AI will be preserving trustworthy human judgment in the face of capable machines.
- [3]
19% of panelists disagree or strongly disagree with the statement, and a further 15% are neutral.
- [4]
The experts diverge on what human judgment means and how it differs from human oversight, and cite its inherent limitations.
- [5]
Beneath the surface, broader agreement appears to exist: the goal is not to preserve human judgment for its own sake but to design human-machine systems that keep trustworthy human judgment available when it matters most.
- [6]
The article says the answer is not simply to keep a human in the loop.
- [7]
Stefan Verhulst, GovLab chief research and development officer, said: 'Human judgment was never inherently trustworthy.'
- [8]
Yan Chow of Automation Anywhere said that 'humans face fatigue, bias, lack of experience or expertise' (the quote continues beyond that point in the source).
- [9]
Yan Chow said human judgment matters most where machines are weak: novel situations, ethical trade-offs, and context not captured in the training of large language models.
- [10]
Ben Dias, IAG's chief AI scientist, said: 'The defining challenge of responsible AI will most certainly be preserving trustworthy human judgment as increasingly capable AI systems operate at greater speed and scale.'
- [11]
New research suggests AI overuse or overreliance risks degrading cognitive and moral judgment, resulting in 'epistemic de-skilling' and reduced critical thinking skills.
- [12]
Evidence-based concerns about AI's impacts on children's cognitive and emotional development led New York City to introduce the nation's first ban on generative AI in schools.
- [13]
The article says it offers practical recommendations for organizations looking to create systems that build in room for meaningful human judgment as AI grows more pervasive and capable.
Get this ledger inside ChatGPT. Ask about any of our stories and the answer brings its figure, each claim’s verdict and the sources. Free, no account.
Sources
1 independent publisher whose own reporting we read for this story.
Topics and entities
Follow any of these and your For You feed starts watching them — no settings page required.
Topics
- AI and cognitionFollow
- Responsible AIFollow
- Human Oversight and Validation of AI OutputFollow
Entities
- MIT Sloan Management ReviewFollow
- Boston Consulting GroupFollow
- Stefan VerhulstFollow
- Yan ChowFollow
- Ben DiasFollow