Time to Write: 5+ hours
I have started (and then restarted) this post several times. After an hour or so of plugging away and feeling frustrated then finding creative new ways to procrastinate (highly recommend Daniel Piper “Serious Literary Author” on Substack for this) and then remembering I have a deadline, I asked Claude to suggest some hooks.
But then felt like a complete hypocrite outsourcing the opening of a post about AI Rot to AI.
One of the things that scares me most about AI (outside of the obvious fears for humanity and the world ending kinda stuff) is what it is doing to my brain. And what it will do to my daughter’s brain.
I have built my career on researching and thinking deeply about different topics relating to how we work. And then writing and speaking about those topics. And turning them into programs that my team and I deliver at Inventium.
Much of my identity is build on these abilities.
But on bad days, I feel the pull of AI in so many insidious ways.
How easy it would be to outsource this post to AI rather than slave away at it for five (very non-consecutive and painful) hours.
How easy it would be to prompt Claude to be running five other tasks in parallel while I write, so that I can get that feeling of insane productivity: “Look at me working while my five agents are also working”.
How easy it would be to stay in denial about the potential to atrophy the precious thinking muscles that we all possess and that I have spent decades building. I worry about how quickly they might waste away if I get too careless about what I give away to the robots.
Now don’t get me wrong: this is not an anti-AI post. I obviously use it in many, many ways, as I write about regularly in this very Substack.
But today I want to highlight what at Inventium.ai we call the Four Horseman of AI Brain Rot. (Dramatic? Yes! Inaccurate? Not at all).
I’ll go deeper into each one over the coming weeks (including what the hell we can do about it), but for now, here they are in a nutshell.
AI Energy Rot
If you’ve been following my work for a while, you’ll know I burnt out (badly) in 2024. Then I wrote a book about it (as you do).
I’m fully recovered now, but I still have days where I am lower on energy (because I am human). And on those days, it is so very tempting to outsource thinking to AI. I suspect I am not alone.
Here are some of the major ways AI is draining our energy:
Attention fragmentation: The constant switching between AI threads, agents and other tasks while you wait for outputs. It breaks your focus and costs time every time you come back. (And we thought task switching was bad before AI…)
Verification burden: The mental effort of checking AI output for errors. This is exhausting because AI sounds more confident that an MBA grad from Harvard, meaning you get none of the usual warning signs that poor quality work used to give us.
Cognitive overload: Having more information, inputs and AI outputs to process than your working memory can handle.
AI fatigue: The weariness that builds from constant AI tool use, new releases and hype (and doom!), leaving people drained and less motivated to engage. Can you remember the last day when there wasn’t a brand new model / transformational new tool / exciting new feature released? I think it was sometime in 2022.
Technostress: Stress caused by the pressure to adapt to, keep up with and always be available through new technologies.
Decision overload: Mental depletion from the extra micro-decisions AI creates, like which tool to use, which output to accept and what to edit, which when we are fatigued, leads to poorer choices over time.
AI Brain Fry: Acute mental exhaustion from intensive AI use, especially from overseeing and switching between several AI tools or agents at once.
AI Skills Rot
Think for a moment about why you were hired into the job you are currently in. Or if you are a business owner or freelancer, reflect on why your clients or customers hire you. I imagine it’s got something to do with your skills.
Perhaps ironically, the tool we are now all using to get more done in less time for whoever hired us, is making us worse at the very thing that made us so hireable in the first place. It’s the equivalent of hiring a personal trainer to build strength, but outsourcing the entire workout to the PT.
Here are some of the ways AI is rotting our skills:
The performance-learning gap: Doing better with AI while learning less, so that people perform worse once the tool is taken away.
Cognitive offloading: Handing mental work such as remembering, calculating or reasoning to AI, which lowers effort in the moment but weakens those abilities over time. (As an aside, how many phone numbers do you know off by heart? If the answer is your parent’s landline from 1994, I see you).
Metacognitive laziness: Letting AI plan, monitor and evaluate your thinking for you, so you reflect less on whether you understand something. Over the long term, this leads to:
Cognitive debt: The build-up of long-term costs to memory, understanding and ownership of work that comes from repeatedly letting AI think for you.
Junior pipeline erosion: Juniors lose the tasks they learn from (first drafts, basic analysis, routine code, pattern recognition) because AI now does them, so fewer people build the expertise needed to catch AI's mistakes.
General deskilling: The gradual loss of expertise and capability when we stop practising skills that AI now does for us. Case in point: you learnt French at school for 5+ years and can now only say Bonjour and Croissant. Not that I am speaking from personal experience.
Homogenisation: AI makes people’s work more creative on its own terms (at least for average and below average performers) but makes everyone's work more alike, so a team or organisation ends up with less variety of ideas.
AI Judgement Rot
Let’s go deeper into one skill in particular: judgment. I think this deserves a category in and of itself. Because without good judgment, well, what exactly do we have left that can’t be replaced by AI?
When I think about my own work, good judgment is behind everything that matters. Designing keynote speeches that resonate and impact audiences. Booking top guests for How I Work (and rejecting the 100+ pitches I receive weekly from guests I don’t believe would be a good fit), selecting topics my audience wants to read about on Substack, designing products that our clients value at Inventium and Inventium.ai. I could go on. And on. But let’s look at how AI is eroding our ability to make good judgment calls.
Automation bias: Favouring AI suggestions and overlooking contradictory information or your own judgement, even when the AI is wrong (and it often is).
Over-reliance: Trusting and depending on AI beyond what its actual accuracy deserves, so errors go unchecked.
Superficial processing: Skimming and accepting AI outputs at face value instead of engaging deeply enough to evaluate or understand them. (Sadly, the internet trained us to do this long before AI came along. And if you have read this far without skimming, I love you very much).
Sycophancy effects: The distortion that occurs when AI tells you what you want to hear, reinforcing your existing views and inflating confidence in flawed ideas. (Many people do not need help growing their egos and manage perfectly fine on their own). And even worse than that is:
Delusions & amplification: AI conversations reinforcing and escalating false or extreme beliefs, sometimes to the point of losing touch with reality.
The productivity illusion: Feeling faster and better with AI when you are not. Research has found that even when we believe Ai is speeding us up, we are often slower (or the same speed).
Persuasive fluency: Being convinced by how well something is written rather than whether it's right. And ultimately:
Reduced critical autonomy: A weakening of your ability and willingness to think independently, question outputs and form your own conclusions. Or, in simple terms: you stop being the boss of your own brain.
AI Behaviour Rot
So how is all this showing up in terms of how we behave? Sadly, in lots of ways:
AI Slop: AI-generated work that looks polished but lacks substance, shifting the effort of fixing or decoding it onto whoever receives it. You know it when you see it: that uber-long email with randomly bolded phrases and way too many em dashes and words like “delve” and quietly” appearing indiscriminately.
Hyper-engagement: Excessive, hard-to-stop use of AI tools that crowds out rest, other activities and human interaction.
AI-driven overwork: Working longer and harder because AI raises expectations and makes it possible to take on more, instead of freeing up time.
Compulsive checking of your bots and agents: The urge to constantly monitor, prompt and manage all your AI agents and outputs, keeping you tethered to your tools. It’s the equivalent of a new parent being glued to the baby monitor (except the baby writes code and can delete your files).
AI dependency: Reaching a point where you feel unable to do tasks, make decisions or feel confident without AI.
Accountability diffusion: No one owning the output because "the AI did it", so errors slip through and blame gets shifted. It’s the 2026 version of “my dog ate my homework” but you are paying a monthly subscription for said dog.
Tell me in the comments: which one can you most relate to? (And if your answer is “all”, I suspect you are in good company).
My team at Inventium are working on programs to overcome these challenges. Keen to put your name on the waitlist to be the first to know when we launch? Take 60 seconds to fill in this form.