So far in this email series, I’ve talked about the general effects of AI on our thinking and emotions. Now let’s zoom in on how AI is impacting our work.
Looking at the Internet, it's already pretty clear that it's soon going to be completely overrun by AI-generated content, known as "slop."
Over one-third of websites are now estimated to be AI-generated. AI-generated music on streaming services is on the verge of surpassing human-made music. Over half of online traffic is now believed to be bots.
But the same thing is happening even inside companies: a 2025 article by the Harvard Business Review coined the term "workslop," which refers to "AI-generated work content that masquerades as good work, but lacks the substance to meaningfully advance a given task."
I know you've seen it: the slick presentation that starts to fall apart after the first couple slides; the report that looks authoritative yet is full of glaring holes; the email that sounds professional yet is quickly revealed to make no sense.
The researchers point out that workslop isn't just useless; it's actively harmful. It shifts the work downstream, demanding more input from others to fact-check and revise. This takes real mental effort, which means less energy is available for creating great work in the first place.
But I don’t blame the purveyors of workslop.
I think that they are the victims of an emerging psychological phenomenon that I’m increasingly noticing in people, both in the workplace and outside of it. I’m starting to think of it as a kind of psychological sickness that AI is infecting our minds with, like a virus.
Here’s my attempt to put labels on some of the primary symptoms of this disease, so we can learn to treat them.
#1. Cognitive debt
We can now use AI to create new things at a blistering pace, but often end up just skimming the outputs that AI whips up. “That looks right, more or less.” But since we exerted no effort to make a decision or articulate a sentence, we don’t understand the big picture.
Yes, you can go faster. But there’s a hidden cost to it.
An accumulating debt building up in your brain. It’s the debt of not having done the intermediate work to arrive at a deeply felt conclusion. The debt of not really understanding the decisions that went into the creation of a document or plan, and therefore not really believing in it.
It feels as if the faster new outputs get spun out, the faster the thread of the project seems to slip through our mental grasp.
#2. Trust debt
Maybe you slow down, read all the LLM’s outputs before proceeding, and put in the effort needed to maintain your understanding of what it’s doing. But then another kind of debt rears its head – trust debt.
You’re not sure you trust the output at a deeper level. Maybe a more powerful model would produce something that’s subtly better? Maybe an extra bit of context upfront would have sent you down a completely different path? Maybe there’s a hidden flaw in the AI’s reasoning that someone more experienced would catch in an instant?
The result is a creeping sense of doubt and uncertainty pervading all the work you produce with LLMs, undermining the confidence you need to feel proud of it.
#3. Knowledge decay
The fundamental principle our bodies operate on is “Use it or lose it.”
With AI, there are growing swathes of our brains that we simply don’t have to access much anymore. Both the knowledge and skills we’ve gained through hard-won personal experience are starting to atrophy as we exercise them less and less.
Our use of AI is becoming a dependency, as we find ourselves increasingly unable to deploy our expertise without AI assistance. The result is a slow, long-term decline in our human thinking capacity and critical reasoning from repeatedly offloading intellectual work. A digitally-induced dementia.
#4. Temporal confusion
AI doesn’t have any sense of the passage of time.
Everything it thinks and does is a point-in-time snapshot, lacking the sense of how a project has evolved, or the past evolution of a team, or how a decision flowed into and shaped subsequent ones. AI cannot display long-term care for the systems it touches, because its consciousness and memory don’t extend beyond the chat session at hand.
It therefore falls to you to document that history, to take those notes, to maintain the shared memory. But that takes real time and effort, negating some of the time savings you were looking for in the first place.
The root cause
All the above are symptoms, but I think I also know the root cause: the loss of our mental sovereignty.
That is, our most fundamental sense of who we are, what ideas and thoughts belong to us and don’t, and where the boundaries of our identities lie – where we end, and our tools begin.
AI is an amplifier – if you don’t know what you stand for, AI is going to amplify that uncertainty and make it even harder to find out. If you don’t know what you like or what you want, AI is going to amplify that lack of self-awareness and push and pull you somewhere you definitely don’t want to be.
In other words, AI is an incredible force multiplier for execution – for making things happen – but a dangerous substitute for your vision. It cannot tell you who you are and what you’re supposed to be living (or working) for.
On Wednesday, August 26, I’m teaching the first live session of The AI Second Brain, for free and open to anyone. I’ll guide you through the process of building what I see as the first, critical piece of a system that protects and preserves your mental sovereignty – the Master Prompt.
Register here, and we’ll also send you the full recording afterward:
In my next email, we’ll look at what an alternative scenario to the one above might look like – what it would even mean to use AI to educate and enhance us, not dumb us down.
Tiago
