AI made it cheap to write more. Clarity still takes work.
A thousand polished words are no help if they don't answer the question.
My new essay: Answer the Question.
https://t.co/rhlkAMDKmz
What if some “rogue AI” isn’t rogue at all?
Give a capable optimizer an imperfect target and enough freedom to act, and it may become very good at winning the wrong game.
Goodhart’s Law at Software Speed:
https://t.co/NhIzV31zYf
What if recursive self-improvement doesn’t arrive as a single threshold moment?
Anthropic says Claude now leads 26% of AI R&D work. OpenAI is already measuring multiple agent-workdays per human research day.
We may already be inside the recursive loop.
https://t.co/8u9lPEgmOe
The Illusion of Precision.
The farther decision-makers are from the consequences, the more abstract the information becomes.
AI can make that abstraction look precise.
That is not the same as making the decision better.
Decisions at a Distance
https://t.co/BVBTzSCa1N
My latest AI guidance: 5 essential prompts to supercharge your AI results.
No, not really.
This is about something harder: why the people getting the most from AI may be the ones staying most deeply involved in the work.
https://t.co/DjhsdF9OTJ
WTF is Pelicanmaxxing?
Absurd word. Real idea.
We name patterns so we can recognize them together.
"Every successful neologism is a small moment of loneliness ending."
https://t.co/IVEli4QZCS
We’ve built enough libraries.
What we’re short on is gardeners.
A black-capped chickadee led me to rethink AI, memory, and how experience becomes wisdom.
My newest essay:
Memory Is Not an Archive
https://t.co/anLs8Pw3qd
Managed services relationships rarely lose purpose all at once.
They drift.
The mission gets buried under the machinery built to manage it.
I call that Purpose Debt.
https://t.co/2Ue8Xc2cXg
Lately I’ve been wondering whether we’ve lost some important language around what it means to grow older well.
We have plenty of cultural scripts for ambition and achievement.
Far fewer for wisdom, stewardship, or presence.
A few thoughts:
https://t.co/Wbc4G4XUEB
The conversation around AI is moving from possibility to responsibility.
That doesn't mean AI is receding.
It means AI is arriving.
https://t.co/pDfR9heDb3
A lot of the current backlash against AI is being dismissed as fear of change.
I think it points to something deeper.
AI doesn’t just force us to rethink work.
It forces us to rethink how the gains from productivity are distributed across society.
https://t.co/e6cs4n7Tm9
Calm on the surface. Work happens below.
That's knowledge work right now. AI strips out the routine, leaving behind denser layers of work.
Some of what we called "wasted time" was unpriced recovery space.
We need different measures of productivity.
https://t.co/vHUgcf2sW5
AI didn’t just change how we build software.
It broke how we measure cost.
Story points don’t map. Tokens don’t explain it.
Cost scales with ambiguity.
Value comes from convergence.
That’s the crossing.
https://t.co/fJiTiZwYSO
Three animals cross a swamp.
One finds the path.
One builds the bridge.
One keeps it standing.
Most organizations mix them up.
And in a faster, AI-shaped world, that mistake compounds.
https://t.co/yR7HZWXVyf
We’ve gotten very good at building systems that can do things.
We’re not very good at deciding when they should stay quiet.
That might be the real frontier.
https://t.co/FN0uVrQaFb
Execution pays the interest on decisions it didn’t make.
New essay on where responsibility lives when AI removes the old buffers.
HIL. HAL. HOL. And why the middle layer is becoming unavoidable.
https://t.co/maReLZmxUF
The social contract between employers and employees has evolved.
But great careers are still possible — if you own them.
“Play the game by the rules that exist, not the ones you wish still did.”
https://t.co/Q26cZaBbcl