35 Years of Judgement Doesn't Fit in a Prompt

There's a quiet promise underneath a lot of AI hype: that judgement is becoming optional. Feed the machine enough data, the story goes, and the right decision falls out of the other end — and all the human stuff, the gut feel, the instinct, the hard-won sense of what'll actually work, is treated as a bug to be engineered out. It's a tidy theory. It's also wrong. The real relationship between **human judgement and AI** is a partnership, not a replacement. I've spent the better part of thirty-five years doing this kind of work, most of it long before any of this technology existed. I use AI every day now, and I'm glad of it. But when a real decision has to be made, the thing I trust most is my own judgement — and that has never fitted in a prompt. I don't think that's nostalgia. I think it's the most honest read of what these tools actually are. ## Your gut isn't sacred either Here's where I'll disappoint anyone hoping this is a "trust your instincts, ignore the data" piece. It isn't. Experience has its own failure mode. It curdles into "we've always done it this way." Instinct can be plain old bias wearing a confident face — the client you rate because they remind you of yourself, the tactic you love because it worked once, a decade ago, in a market that's since moved on. The whole first half of this series exists because gut on its own leaves money unmade and gets things quietly wrong; because a business that never checks its hunches against the evidence is guessing just as surely as one that automates without thinking, as in [You Can't Automate Guesswork in AI Marketing](/insights/ai-marketing-cant-automate-guesswork/). So this isn't experience versus data. Anyone selling you that fight is selling you half a brain. ## What experience gives you that a model can't What experience gives you — and no amount of data quite replaces it — is narrower and more valuable than being right. It's knowing when the confident answer is wrong. An AI is fluent. It'll hand you a clean, plausible, well-structured answer to almost anything, in the same assured tone whether it's right or nonsense. Experience is what reads that answer and thinks: that doesn't smell right. You can't always say why in the moment — it's every time you've been burned before, quietly adding up — but it's real, and it's the difference between catching the confident mistake and shipping it. It's the same instinct that decides [which jobs are worth handing to AI](/insights/jobs-worth-handing-to-ai/) in the first place. Experience also tells you which question to ask in the first place, which AI never will. It answers the question you give it; it has no idea whether it's the right one. Knowing you're about to solve the wrong problem is a very human skill, and you usually learn it the hard way. And then there's everything that never made it into any dataset, because it only ever lived in your head. That this client always says yes and then goes quiet. That when this supplier says "no problem," it's about to be a problem. That this kind of job always overruns, whatever the quote says. That's your proprietary data, and it isn't written down anywhere a model could reach it. ## Not gut versus data — gut and data So the answer isn't to pick a side. It's to make them argue. Use the data, and the AI, to interrogate your instinct: I think this client's our best fit — does the evidence actually back that, or do I just like them? And use your instinct to push back on the data: this analysis says go all-in on that segment — does that square with what I know about how those jobs really go? When the two agree, you can move quickly and with real confidence. When they disagree, you've found the most useful thing in the whole process — a place to look harder. That disagreement is a gift, and it's one a purely gut-run business and a purely data-run business both miss entirely: one never hears the challenge, the other never feels the doubt. That's the right relationship with these tools. AI is a fast, tireless, occasionally brilliant analyst that has no judgement and carries no accountability. You have both. It does the legwork; you make the call. Get that the wrong way round — letting the confident tool make the call while you do the legwork — and you've handed the one part that was actually yours to something that can't be held responsible for it. ## The thing that doesn't fit in a prompt So if you've been quietly wondering whether decades of experience still count for much in an age of clever machines, here's the honest answer: they count for more, not less. When answers are cheap and confidence is free, the rare and valuable skill is the judgement to know which confident answer is wrong. That was always the job. AI just raises the price of not having it. It's why our [AI Strategy & Roadmap](/services/ai-strategy-roadmap/) work keeps a human firmly in charge of the call. The verdict on this one: AI is support, not replacement. Your judgement isn't what's being made obsolete — it's what's doing the deciding, now with a very capable assistant to do the reading. Keep it that way round and you get the best of both. Swap them over, and you've automated away the only part that was ever really yours. ## Frequently asked questions ### Is human judgement still relevant now that AI is so capable? More than ever. When confident answers are cheap and free, the rare skill is knowing which confident answer is wrong. AI is fluent whether it's right or nonsense; experience is what catches the plausible mistake before it ships. ### Should I trust my gut over the data? Neither on its own. Use data and AI to interrogate your instinct, and use your instinct to push back on the data. When they agree, move fast; when they disagree, you've found the most useful thing in the process — a place to look harder. ### What's the right division of labour between AI and a decision-maker? AI does the legwork — reading, analysing, drafting — and you make the call and carry the accountability. Reverse that, letting a tool that can't be held responsible make the decision while you do the busywork, and you've handed over the one part that was actually yours. If you're ready to understand where AI can genuinely save you time, reduce friction and create real capacity in your business — without automating a guess — we're happy to talk it through. [Book a discovery call](/contact/) and we'll look at your situation honestly. *Peter Lowe — AI Consultant and Founder, Smart AI Studio.*

This article was written by Peter Lowe. The ideas and opinions are his own; AI was used to assist with drafting and editing.