Human-First AI Pulse

Two new studies put a number on something a lot of people feel but haven't named. BCG surveyed nearly 12,000 employees, and Glean's Work AI Institute found workers now spend about 6.4 hours a week supervising and correcting AI output (Wieczorek, 2026). Most get no guidance at all on what to do with the time AI supposedly saves. The hours don't disappear. They move. If AI joined your business this year, the honest question isn't whether it saves time. It's who's checking its work, and what that checking costs them.

A Getsolved survey of 3,000 professionals under 30 (industry data, worth naming as such) surfaced a contradiction worth sitting with (Getsolved, 2026). 75 percent say AI makes them more productive. More than half have still skipped using it because the oversight felt too draining, and most employers aren't tracking that cost at all. Here's the humanity-first read: what looks like resistance is often energy protection. These aren't people afraid of the future. They're people whose energy is already spoken for, making a rational call. Adoption plans that ignore that will keep mistaking self-preservation for stubbornness.

The American Psychological Association's 2026 Chatbots and Mental Health Survey is the clearest primary-source look yet at how AI is entering care (American Psychological Association, 2026). 77 percent of psychologists have talked with patients who use AI for emotional support. A third say patients treat it as an added therapy tool. Meanwhile, 94 percent of psychologists say they don't trust tech companies with that mental health data. AI isn't arriving in helping professions through policy. It's arriving through the people being helped. The practitioners in the room deserve a plan that catches up to what their clients are already doing.

JumpCloud's Q3 2026 IT Trends Report (vendor data, widely reported) tracked a sharp slide in six months: organizations requiring human review before high-risk AI actions dropped from 40 percent to 25 percent, while fully autonomous AI with no human review more than doubled (JumpCloud, 2026). Non-human identities now outnumber human users in 83 percent of organizations, yet only 21 percent have governance controls for them. Trust that grows faster than oversight isn't confidence. It's exposure. "We adopted AI" and "we're governing AI well" are two different claims, and the gap between them is where trouble starts.

New research in Harvard Business Review on AI adoption in modular firms found the constraint usually isn't the tools. It's outdated governance and unclear decision rights (Harvard Business Review, 2026). Companies can break work into pieces faster than they can put it back together with people and trust intact. Before adding another AI tool, the more useful move is naming who actually has the authority to make decisions with it. That's usually the real blocker. The technology was never the hard part. Holding the humans together while it changes their work is.