She Reviewed Forty-Three Applications Before Lunch. She Was Allowed to Question None of Them.
SpatialNext
Sarah is a recruiter with eleven years of judgment and an AI that hands her a ranked shortlist she has no real authority to question. When one candidate’s strong record lands below thinner profiles, she sees the problem — and moves it forward anyway, because nobody designed what she’s allowed to do when she disagrees with the machine. The piece traces that gap through the Mobley v. Workday class action — where 1.1 billion applications were screened and one rejection arrived at 1:50 a.m. with no human awake — and the new wave of law, led by Colorado, now demanding the one thing most organisations never built: a human with the actual authority, training, and protection to overrule the model.
🔗 https://spatialnext.io/2026/06/29/she-reviewed-forty-three-applications-before-lunch-she-was-allowed-to-question-none-of-them/
When humans and AI work best together — and when each is better alone
MIT Sloan / Nature Human Behaviour, 2026
The largest study of its kind looked at over 100 experiments and found something that should stop every “human-in-the-loop” assumption cold: on average, human-AI teams made worse decisions than AI alone — and the losses were concentrated precisely in decision-making tasks. Adding a human doesn’t automatically make the decision better. Sometimes it makes it worse. Which is the whole question I’m chasing: not whether there’s a human in the loop, but whether the pairing was actually designed to decide better than either side alone.
🔗 https://mitsloan.mit.edu/ideas-made-to-matter/when-humans-and-ai-work-best-together-and-when-each-better-alone
Deskilling Dilemma: Brain Over Automation
Frontiers in Medicine, 2026
This is the clearest primer I’ve found on the failure mode nobody likes to talk about: what happens to a skilled professional who leans on AI for too long. It names the mechanisms — automation bias, where you defer to the system even when it’s wrong; deskilling, where your own judgment quietly atrophies from disuse; and the more troubling “never-skilling,” where the expertise never develops at all because the machine was always there. Written for medicine, but the pattern travels to any field where a human works alongside a confident system.
🔗 https://pmc.ncbi.nlm.nih.gov/articles/PMC12909220/
AI and the Future of Human Decision-Making
Deloitte, 2026 Global Human Capital Trends
Deloitte’s read on where this is heading: 60% of executives now use AI to support their decisions, and oversight is racing to keep up. It’s a serious, data-backed look at human agency in an AI-driven organisation, and worth your time. My one push-back is where it lives — the boardroom. The harder version of the same problem is out in the field, where the call is physical, fast, and can’t be taken back.
🔗 https://www.deloitte.com/us/en/insights/topics/talent/human-capital-trends/2026/decision-making-with-ai.html
Most organisations deploying AI have checked whether the model is accurate. Far fewer have checked whether the human and the machine, together, actually decide better than either would alone — especially when the call is physical, fast, and can’t be undone.
If that question is live in your world, it’s usually where my conversations start: mattsheehan@spatialnext.io