You have a real situation. The operations lead who agreed to your pilot twice and has moved on it zero times. The category manager who nods in every meeting.
So you type it in. Two seconds later you have five headers, a tidy list, and a closing line about the importance of stakeholder alignment. You read it and think: yes, I know all of this.
Which is not the same as knowing what to do on Thursday.
ChatGPT answers instantly. That is the problem.
It is a fair question
General AI is good at a startling number of things, it is already open in your other tab, and it costs a fraction of anything else you could reach for. Most of the people I talk to tried it first.
The ones who call it useless usually mean something more precise: it gave them advice they could not use on a specific person.
What generic AI gives you
I asked mainstream AI how to lead change and got eight answers in under a second. Not one of them holds up when you are leading change against the odds.
→ "Run an awareness campaign so people understand why this matters." If awareness worked, nobody would smoke, skip recycling, or leave the lights on.
→ "Model the behavior yourself and others will follow." If that were true we would all be do-gooders by now.
→ "Build a strong business case so the logic is undeniable." Hypothesized returns don't change minds when the rollout isn't feasible.
→ "Host a training session to explain the technicalities." Training doesn't change anyone's KPIs or workflows.
→ "Get sign-off from leadership so the initiative has real authority." Top-down only works when the reality on the ground is receptive.
→ "Put together a detailed report so the data speaks for itself." Reports don't move people who have already made up their minds.
→ "Make sure you're meeting the compliance requirements." Compliance gets you disclosures, not transformation.
→ "Find the people who care most and put them in charge." Most people already care. Without influence it wears thin.
These sound like good advice, and that is exactly the problem. General-purpose AI reflects the most common thinking on the internet back at you, and the most common thinking about change is that people act once they understand. Behavior science has spent decades finding out that they don't.
The issue is sequence, not quality
Every answer above is an answer. Which means the model decided what your problem was before it knew anything about the people in it.
Ask a specialist and the first thing you get is a question. What has this person already tried. What does the change cost them personally, in time, in risk, in looking foolish in front of their own team. Who do they actually listen to, which is almost never the person above them on the org chart.
Instant is the wrong shape for this work. The diagnosis is the work.
What a specialized thought partner does differently
It refuses to answer until it has found the actual barrier. It asks about the specific human, not the initiative. It makes you consider options you would not have reached for, because the first one that occurs to any of us is usually a version of explain it again, better.
Then it gives you something you can carry into the room: a written plan with when, where, who supports it, and what could derail it. Afterwards it checks in by email, because a plan that lives only in a chat window is a plan that quietly expires. And when a situation is a real crisis rather than a stuck conversation, it hands you to a human.
That is Real Change Partner, built on 25 years of behavior and implementation science.
When ChatGPT is the right tool, and it often is
Be generous with it. It is good at drafting the supplier email once you know what the email needs to accomplish. At compressing a sixty-page directive into something you can act on. At turning messy notes into a slide outline, or rewriting a paragraph for a CFO who hates adjectives.
Use it for everything downstream of the decision.
Just don't ask it to make the decision about a person it has never asked a single question about. Knowing more has never been the thing that makes people act.
One question to sit with
Think about the stuck conversation that has been sitting on your list for three weeks.
What would you need to find out about that person before any advice about them could be worth having? And has anything you have typed into a chat window so far asked you for it?