There is a strange phenomenon happening right now in the world of technology.
SPECIFICALLY WITH AI…
The smarter AI gets, the more exhausted we feel using it.
If you've been feeling burnt out trying to keep up with the latest "prompts" or struggling to manage your messy digital life, you are not alone.
In fact, some of the top minds in the world are feeling the exact same way.
This week, I watched three fascinating discussions that perfectly capture why we are all feeling so overwhelmed by AI right now…
1. The "Exhaustion" of Unlimited Potential
In a recent conversation on the No Priors podcast, Andrej Karpathy (former Director of AI at Tesla) talked about the overwhelming “cognitive load” of managing AI agents.
He described a state of constant anxiety… I’ve felt myself.
The feeling that because the AI can do so much, if you aren't perfectly utilizing it 16 hours a day, you are falling behind.
When AI has unlimited potential, YOU (as the human operator) become the bottleneck.
2. The Unpredictable "Black Box"
Why is managing this AI so stressful?
Because, as a recent deep-dive by Petr Lebedev highlighted, AI models are grown.
(Not “programmed” like a regular computer.)
Even the researchers who built AI don't fully understand how it decides to output specific words.
When you try to have an open-ended conversation with a system like that, it's inherently unpredictable.
We all know AI often gets distracted, loses context, and hallucinates.
(From my perspective, the current AI structures we use are inherently designed to to always produce these as results.)
3. The Antidote: AI as "Normal Technology"
So, how do we regain our sanity?
In a great episode of Factually, researchers Arvind Narayanan and Sayash Kapoor argue that we need to stop treating AI like a mysterious sci-fi entity and start treating it like normal technology.
This is exactly what we have been quietly building in the background.
I believe the solution to AI exhaustion is boundaries.
You don't need a better prompt; you need a better digital desk…
A better way of ORGANIZING with AI.
If you lock your AI into a strict "Workspace" with physical folders (an Inbox for messy ideas, a Rulebook for instructions, and an Outbox for finished work), it dramatically reduces the chances of confusion, drift, or hallucinations.
It stops acting like some conversational chatbot…
And it starts acting like a predictable, reliable, automated thinking machine.
This Wednesday I’ll send you some examples.
In the meantime, if you’re comfortable with GitHub and want to tinker…
If you’re familiar with GitHub already, it should only take a couple of minutes to set up in Gemini or ChatGPT, zero coding required.
Either way, I’ll send you some more interactive “LEAP” experiences this Wednesday!
Thanks for being a part of the #ThisIsYourMindOnAI newsletter!
— Mercer