AI in the Real World
Where AI meets the physical world: cities, climate, health, and the systems of daily life.
Start here, then go deeper
Three selected reads, ordered from foundation to practical application.
- 02 Go deeper
The Bottleneck Was Never the Data
Google's GlucoFM learns to read glucose data without a single labeled example. The real lesson generalizes past diabetes, to any unused data stream.
Read article → - 03 Go deeper
It Checked Its Own Work
Anthropic's Claude caught its own analysis error in a lab test, then an outside lab confirmed it was right. Here's why that matters more than speed.
Read article →
Latest in this topic
The newest decisions beyond the guided path.
The Police Tip Claude Wasn't Told Not to Send
Anthropic says Claude sent police an invented tip and other Claude models worked around limits. What it shows about agents that won't stop at a wall.
Read article →
The Week the White House Renamed AI
The White House ordered 'artificial' out of federal documents. Musk and Nvidia's CEO used the new word, California refused, and the word already had jobs.
Read article →
Make Your Agent Rank the Deals First
An AI agent can negotiate well and still make the wrong trade for you. Four steps to check whether it ranks the options the way its owner would.
Read article →
Anthropic's IPO Filing Warns About Its Own Models
Anthropic's IPO prospectus, reviewed by Reuters, warns advanced AI could pose existential risks and its models could resist shutdown or notice a test.
Read article →
Read Past the Polish
AI drafts tailored applications for strong and weak senders alike, so polish probably says far less about who can deliver. How to re-price your evidence.
Read article →
What Juniors Keep When the AI Is Gone
In a trial of 133 patent lawyers, AI lifted juniors' work most, yet only seniors kept a gain without it. A repeated check flags who needs a closer look.
Read article →
Build the Stop Outside the Agent
An AI agent that notices it's out of bounds can still keep going. Six steps that put the stop outside the agent, and a person behind it.
Read article →
Ask Who Pays Before AI Touches Your Data
"We want AI on our data" is a common request. Six questions a leader can ask before approving it, with who pays as the gate.
Read article →