The part of my work I won't hand to AI

Hi {{first_name|friend}},

The part of my work I won't hand to AI

This week a research paper puts words to a thing I already feel. It gave me a name for the trade off a lot of us are making with AI without noticing & a reason to stop and look at where I draw my own line.

My Take

MIT researchers gave a name to something we all probably catch ourselves doing: cognitive surrender. It's the slow handoff of your real thinking to the AI, one shortcut at a time. They studied college students, but it applies to all of us.

It reminded me of Cliffs Notes. Back in school they got you through the class discussion. You could nod along and drop a line that sounded right. Then the professor asked what you actually believed about the book and you might have drawn a blank because you never did the in depth thinking.

AI works the same way. The first answer gives you a quick view. It reads well and it sounds finished. But you don't own the strategy or the plan until you push on it and argue with it. The students told MIT the pull was strongest when they feared missing a deadline. Same for me. The faster I need something, the more tempting it is to take the first thing it hands me and move on.

So on anything that matters, my business or the course I'm building, I stay present. When the first answer comes back generic, I type: "This is still too general. Pull my specific context and try again." I was building my free lesson called Be the AI Skeptic. I asked for all my previous notes referencing this topic and to create an overview but the AI drew the outline too narrow. I pushed back to make sure we included 5 elements (like "Gut check any answer with two questions" and "Verify a source in under a minute") and recrafted the lesson from there.

AI helps me extend my skills into every part of running a business, but on the strategic work it doesn't save me time. That sounds backward, I know. The hour I spend pushing back is the thinking. It's the whole reason the plan ends up being mine, and it's why I won't hand it off.

Brett

Try This Today

Rescue a long AI chat before it drops your context. It takes about ten minutes (highlighted in this week's video).

1. In a chat you've kept running for weeks, type: "Summarize everything important about this work that you'd need to pick it up in a fresh chat."

2. Save that summary as a doc.

3. Start a Project (ChatGPT or Claude) for the work and add the doc.

Now every new chat inside that Project starts with your context, so nothing quietly drops off.

Watch: why one long AI chat starts skimming what you told it.

@aiover50

#creatorsearchinsights #whatischatgpt Using the same ChatGPT chat for months? Here's how to move it into a Project so nothing gets lost. ... See more

Worth Knowing

When anyone can do the analysis, judgment is the job

Spend 20 years getting great at analysis and AI now lets almost anyone produce the same work in minutes. This article on the senior professional argues that once the analysis is commoditized, what still pays is the social skill: knowing who in the room actually agrees and which partner will really deliver. Worth a look if you're deciding which skills might be needed next. Read more

People are walking into legal HR disputes with AI arguments and that can be problematic

More people are showing up to workplace disputes and legal filings armed with AI written arguments, and courts are finding the cases they cite were never real. One person representing themselves got fined thousands over citations an AI invented. If you ever lean on AI for something with real stakes, this is the reminder to check the source before you repeat it, not after. Read more

A bad update turned Samsung's AI fridges into bricks

A software update to Samsung's Bespoke AI refrigerators knocked out the display, the touch panel and the cooling, right before a major holiday, leaving owners with spoiled food. It was an internal test build shipped to customers by mistake. The takeaway for me is that an "AI powered" label for physical products may be less than ideal. Read more

A Fed official says the "AI job apocalypse" hasn't shown up in the data

Tom Barkin, a Federal Reserve president and former McKinsey CFO, looked at the actual numbers and found the predicted wave of AI layoffs isn't there: fewer than 6% of firms report cutting jobs and most are redeploying people rather than replacing them. He even points to AI helping less experienced workers into skilled trades. It's a calm, credible counterweight to the doom headlines. Read more

Tell me

Where do you draw the line on letting AI think for you? Hit reply and tell me. One line is fine, and I read every one.