The real slop: confidence without competence

Earlier this year, I ran a workshop for thirty startup founders.

Some people in the room were at the idea stage. Some had prototypes. Some were already live, with apps sitting in the App Store, betas in the hands of customers, dollars starting to change hands and contracts being signed.

One founder told me about his product. It was 100% built with AI and live in the App Store, with ambitions of reaching hundreds of thousands of people. He took great pride in it, and he had earned it with no engineering background. He never wrote a line of code.

I haven’t either.

I asked him a question I find myself asking a lot now: “Have you had an engineer sit down with you?”

“No.”

Not one code review. Not one architecture conversation. Not one hour with someone who has built for scale, for security, for reliability. And to be fair to him, why would he? The app worked. The store approved it. Users were using it.

I also have never written a line of code in my life, and I now build apps I use every week. But I also know those apps would not survive 1,000 real users. Not because I can see the flaws. I cannot. Because I have just enough self-awareness to know there are things I do not know.

So my question to him was not a “gotcha.” I genuinely wanted to know. Why didn’t his lack of coding experience hold him back? No one questioned him. Not the tools. Not the store. Not the users.

The app got approved. Every gate that exists validated his confidence. 

But not one interrogated his competence.

The app works fine now. It’s working for hundreds of people. But what about 1,000? 10,000? If no engineer has ever looked under the hood, a confident output on its own won’t stand the test of time. Confidence must be built on top of a foundation of competence. 

Let's be clear about what this is not.

It is not a tools problem.

It is not a laziness problem.

And it is certainly not an intelligence problem.

This is a signal problem. The output looks finished either way.

For most of our working lives, the output proved the work. If the writing existed, the thinking had happened. If the code ran, someone had understood the problem. We did the work by hand, so seeing the artifact meant the decisions had been made.

That link between the output and the work is now broken. 

This isn’t just apps. The same thing is happening with landing pages, campaigns, dashboards, decks, thought leadership writing. It’s happening anywhere output can be generated.

I wrote about this in a previous edition: the artifact no longer proves the thinking. The artifact stopped proving that the person who made it did the thinking.

And yet, you still get confidence from producing the output. You made a thing. The thing exists, and it looks good, and it works.

Competence, however, lives somewhere else. It lives in the thinking that the output does not record: what was considered, what was ruled out, what should have been feared, what was learned along the way and folded back in.

The two used to move together, because producing anything forced the thinking. Now they are no longer automatically connected. 

AI gave you the confidence. It did not give you the competence.

And here is the uncomfortable economics of it. You can subscribe to the confidence for $20 a month, from a galaxy of tools, with more hitting your inbox every morning. Competence cannot be subscribed to. It has to be earned. 

Your earned expertise and experience still matter. Your ability to interrogate the output still matters, your ability to ask and answer the question, “Is this actually good?”

Last edition, I argued that there is a new job at every desk now: the investor's job, allocating your time, energy, attention, and imagination like capital. And here is where that job shows up next. Not your skills. We have always invested in those: school, degrees, certifications, the training sessions that fill your calendar every year. Your thinking. That is the investment almost nobody has ever been asked to make. Most of us have spent our working lives being paid to comply, not to think.

If we continue to comply in today’s AI-shaped world, our confidence compounds. But our competence atrophies.

In May, the CEO of Shopify, Tobi Lütke, said something I keep going back to: "Everyone is a 7 out of 10 on every skill now."

I have enormous respect for Lütke. He made the statement about his teams at Shopify, where an agentic harness routes expert-grade customer research to everyone automatically. The skill level “seven” is something the harness delivers to the person.

Harness or not, every leader who' has handed out AI tools is now quoting the seven.

Programmers, Lütke said, massively underestimate how much of the work is still theirs when they steer an AI: they have seen the movie before, they can spot the model going down the wrong path, and one or two words from them change everything. His verdict about what happens to people when the tools’ novelty wears off? Everyone falls back onto their own level.

I also think it’s important to notice that Lütke used the word “skill.” Because harnesses can route a skill to every desk in the building. They cannot route the thinking.

The seven describes the output. It says nothing about the competence underneath.

The world already has a name for the gap between the confidence you bought and the competence you have not yet earned. You have no doubt seen it flood your digital world. Eloquent posts with nothing under them. Well-structured reports that dissolve at the first hard question. Apps that work until people actually use them.

The world calls it “AI slop.”

That’s the wrong name for it. 

AI does not wake up in the morning and decide to make slop. A human directed it. A human looked at the output and accepted it. A human put it into the world. Calling it AI slop hands your accountability to a tool and surrenders the one thing you cannot afford to surrender right now, which is agency. If you want any place in the conversation about your own future with this technology, accountability and agency are the price of entry.

And this is not a word game. Vocabulary carries orientation. How you name the problem decides whether you own it.

It is not AI slop.

It is human slop.

The world has noticed the flood, and it is responding. 

At the end of July, LinkedIn shipped a button that lets you flag a post: seems like AI slop. Yet in the announcement, their head of product conceded that "AI and slop are not the same thing", since plenty of people use AI to sharpen thinking that is entirely theirs.

Anthropic has begun watermarking text produced by its models. However, its own documentation is honest about its limits: a detected mark shows content was processed by the AI, not written by it. If you have proofread one paragraph of something you sweated over for a week, then you carry the same watermark as the output somebody generated from three words. The watermark will become a badge of shame, however little or however much you used the tool.

out

Every detector out there  points at the artifact. And a single artifact cannot prove the thinking either way. Yours might be soaked in it. The next one might have none. They read the same. Remember: the artifact stopped proving the thinking. Now watch the world build machinery on the assumption that it still does.

A friend of mine who runs a brand agency put it in seven words on a call last week: just because it works doesn't mean it's good.

Only you can know: did you do the thinking?

You know whether you interrogated the output or just accepted it. That much is visible to you, and owning it is where accountability starts.

Here is what you cannot know from the inside: whether it was enough. The gap is measured by questions you do not know to ask.

LinkedIn's answer is to hand the verdict to a crowd. Strangers, voting on whether your thinking happened, by looking at the one place it can never be seen. That is not detection. And the detector that matters is not theirs to build anyway. It is the one you do not have, pointed at your own work.

I think about this for myself, too.

My videos that I publish on LinkedIn are raw and unedited on purpose. Stream of consciousness, no script, no polish. I do that because rawness is the only proof of thinking that no tool can counterfeit yet.

The written posts that ship with those videos are drafted with AI. In my voice, from my ideas, carrying my thinking, but drafted with AI, and I have never hidden that.

Video and post go out together. One piece of work. If the detectors had their say, the text would carry the watermark and the footage would carry nothing. One post, two verdicts, and both of them wrong about me, in opposite directions, at the same time.

So what do I do? Stop? Go video only? Am I no longer allowed to command a tool to write in my own voice, infused with my own thinking, because the watermark cannot tell commanding from surrender? It cannot see accountability. It cannot see agency.

I do not have clean answers. I am sitting with the same question this edition is asking you.

Your report earns applause in the meeting. It looks polished and complete. Then your boss asks a question. You handle it. She asks a second. You reach. She asks the third why, and there is nothing under your feet.

The confidence compounds. The competence atrophies.

If you lead an organization, sit with this a moment longer. You armed every desk in the building with AI superpowers. What comes back to you is activity. And activity says nothing about competence.

We have already watched what happens when an entire company reads activity as value. Last time it was tokens. This time, it is competence.

For now, the shift is simply this: stop asking what you can produce, and start asking, “Why am I confident this is good?” 

Some questions to sit with, alone or with your team:

The tools are real. The new capabilities are real. Push on them. Build things you could not build a year ago. That part is not in question.

But at the same time, our everyday work is going to be closer to art now. It has to be ours. Of our own making, of our own opinion, of our own perspective. Ours.

Do not confuse the confidence with the competence.

The confidence came with the subscription.

The competence still has to be earned.

That's what I'm watching for Tomorrow.

– Nish

On next week's LinkedIn Live, Wednesday, September 2 at 12:00 PM ET:

A gratitude P.S.

This is me and Tomen on one of our morning walks.

He was a street dog in Mexico City, hit by a car, paralyzed.

A friend's rescue in Phoenix called, and we said yes. I walk with a cane. He rolls on wheels. We both move a little differently.

Some mornings that is the whole lesson: show up, move forward, at whatever pace is yours. My life has been filled with persistent peace and contentment since I started playing my own game.

It took me years to learn that acceptance. Tomen and dogs understand and accept immediately. Disability and dogs. Two of my greatest teachers.