The TLDR Trap and a Drifting Brain
I have so many interests and have ended up subscribing to hundreds of newsletters. So naturally, I thought: why not let AI filter it all for me?
Let's unpack: I maintain an interests.md with a list of topics: e.g. AI Evaluation Tools and frameworks, open-source local inference engine, macroeconomy trend, workforce and education development. Then I created a workflow (first in OpenClaw, then moved to Claude cowork for more reliability) to triage my inbox automatically:
- Read interests and match each email
- If matched, summarize with 2-3 sentences
- Suggest potential new interests to me from irrelevant list
I even layered in Smart Brevity formatting — a one-sentence rundown, context bullets, and a Why It Matters hook. If the TLDR earns my attention, I read the full piece.
Genius, right? I was so excited because it would save me enormous time to filter out content I really care about, so I have more time for deep read.
How did it go?
Yes, it did save me some time, to the point where I never opened my newsletter inbox anymore. And I still find useful information like the frontier model release and why it matters.
For a while, that was enough. But I'm slowly losing the fun of reading.
The funny thing about the human brain is that it needs a hook. That hook is very personal, and I clearly did not understand myself enough to write the best interests list. Sometimes the best finds are buried in the discarded piles. I always feel like I'm on a treasure hunt. TikTok's algorithm does it better than me.
And that's just the matching problem.
The summary may sound smart, but it misses the stuff that matters. The LLM doesn't get what I actually care about — it smooths everything into something readable and loses the weird, specific details that would have caught my eye.
And even when it gets the facts right, it can't tell my signal from my noise. That's personal, it shifts over time, and I'm not sure generic "memory" features will fix it.
But say the AI learns me perfectly — there's a deeper problem. Reading is the thinking. Following a thread, hitting a dead end, backtracking, suddenly seeing the point--that whole messy process is what actually sticks in my brain. The struggle is where my brain connects the dots and builds something lasting.
It's like strength training — the load tears your muscle fibers apart, then satellite cells kick in, fuse into the damage, and rebuild everything thicker and stronger. No amount of watching someone else lift will trigger that process. You have to feel the strain yourself.
I still find my workflow useful--I just have to stay the one doing the fun and thinking.