How do you keep your judgment sharp when using AI at work?
Make your own call before you look at the AI's. That's the core practice Harry Laplanche, Head of AI Strategy and Transformation at Panasonic, describes to host Andrew Barry on The Learning Culture Podcast. His argument is that when AI can gather the evidence, run the analysis, and produce a recommendation, judgment is the one thing left that you can't afford to outsource, and the only way to keep it is to keep exercising it.
From the episode
#127 - How to Protect Your Judgment When AI Does More of the Work with Harry Laplanche
Sep 8, 2026 · 44 min
What's left when the machine does the analysis
The episode starts from an uncomfortable question. If an AI system can pull together the evidence, analyse it, and tell you what to do, what exactly is your contribution? Harry's answer is judgment: the capacity to weigh a recommendation against context the model doesn't have, to notice when the framing is wrong, and to own the decision.
He's clear that this isn't a sentimental defence of human involvement. It's practical. A recommendation without judgment behind it is a liability, because nobody has actually thought about whether it fits the situation. And judgment, he argues, is a skill that decays if you stop using it, which is exactly what happens when people start accepting AI output by default.
Decide first, then let the AI show its hand
Harry's most concrete practice is a workflow he uses himself. His AI setup gathers all the evidence on a question and lays it out, but it withholds its recommendation until he has written down his own call. Only then does it reveal what it would have suggested, so he can compare the two.
The host describes this as building judgment like a muscle. Every decision becomes a rep: you commit to a view, you see where the model agreed or disagreed, and you learn something about your own blind spots or the model's. Over time you get a calibrated sense of when to trust the AI and when to override it. Skip the first step, and you get none of that. You just get faster at agreeing.
Why access to tools doesn't produce adoption
A second thread in the conversation is why so many corporate AI rollouts stall. Harry's background is in consumer insights, psychology, and behavioural science, and he sees the problem as behavioural rather than technical. Handing people a licence and a training video changes almost nothing.
What changes behaviour is connecting the tool to the work people are actually measured and rewarded on. If using AI well helps someone hit the number their bonus depends on, adoption follows. If it's a side project nobody's evaluated against, it dies quietly. That framing also protects judgment: when AI is tied to real outcomes, people care whether the recommendation is right, not just whether they used the tool.
"Giving people access to tools is nowhere near enough. Adoption depends on connecting AI to the real work people are measured and rewarded on."
The missing middle between vision and daily decisions
Harry also talks about what he calls the missing middle. Leadership sets a transformation vision at the top. Individual employees make hundreds of small decisions at the bottom. The gap in between, where the vision would have to translate into concrete changes in how people decide things day to day, is where most transformations go missing.
His view is that change has to run top-down and bottom-up at the same time. Leaders set direction and remove obstacles, while teams experiment with real tasks and feed back what works. The judgment-first decision workflow is an example of a bottom-up practice that any individual can adopt tomorrow without waiting for a company-wide programme.
What to remember
- Judgment is the one contribution AI can't replace, and it atrophies if you stop exercising it.
- Write down your own decision before looking at the AI's recommendation, then compare the two.
- Treat every comparison as a rep that calibrates when to trust the model and when to override it.
- Adoption comes from tying AI to the work people are measured and rewarded on, not from access alone.
- Bridge the gap between transformation vision and daily decisions with simultaneous top-down and bottom-up change.
People also ask
Does using AI make people worse at decision-making?
It can, if they accept recommendations by default. Harry's point is that judgment is a muscle, and the fix is to keep making your own call before you see the AI's.
How do you set up an AI workflow that protects judgment?
Configure it to present the evidence but hold back its recommendation until you've recorded your own decision, then reveal and compare. That's the workflow Harry uses himself.
Why do AI rollouts fail even when everyone has the tools?
Because access isn't adoption. Harry says people adopt AI when it's connected to the real work they're evaluated and paid on.
Based on Episode 127, "How to Protect Your Judgment When AI Does More of the Work," with Harry Laplanche, released September 8, 2026 on The Learning Culture Podcast, hosted by Andrew Barry. The quote is taken from the episode's show notes.
