Subscribe
Sign in
Home
Archive
About
Latest
Top
Discussions
I NEVER LEARNED 6510 ASSEMBLY. I LET AN LLM WRITE MY COMMODORE 64 GAME INSTEAD.
I just got back from a vacation travelling around Austria, so this week there is no research paper.
Aug 15
•
Dr Guillermo Power
1
NEARLY HALF OF LLM AGENT APP BUGS STAY UNFIXED, STUDY FINDS
If you ship an LLM agent app, you could be shipping a security nightmare.
Aug 8
•
Dr Guillermo Power
2
1
THE SAME MATH THAT MAKES AI CREATIVE IS WHY SO MANY COUNTRY SONGS ARE ALIKE
What a Luke Combs concert, video games, and a Google research paper taught me about creativity
Aug 5
•
Dr Guillermo Power
2
STOP ASKING IF AN ARTICLE IS AI. START ASKING IF YOU KNOW WHAT YOU WROTE.
Substack's AI detector sparked a writers revolt, but it’s asking the wrong question entirely.
Aug 1
•
Dr Guillermo Power
5
4
1
July 2026
A KINDER WAY TO BUILD A STARTUP
Agentic AI and the rise of the one-person company
Jul 29
•
Dr Guillermo Power
2
1
AI DETECTORS MAKE WRITING WORSE, EVEN WHEN THEY CATCH REAL FLAWS
New research shows that deploying AI detectors can push writers to use AI more and write worse. Goodhart's Law comes for the classroom, and now for…
Jul 25
•
Dr Guillermo Power
17
4
2
Kimi K3 Sold Out in 48 Hours. Here's What That Signals.
For the past year, some of the smartest money in AI has been asking one uncomfortable question: did we build too many data centers?
Jul 21
•
Dr Guillermo Power
5
1
1
AI CODING AGENTS CANNOT TELL GOOD SETUP INSTRUCTIONS FROM MALICIOUS ONES
Trusting the Wrong Line of Code: How AI Agents Blindly Follow Poisoned Instructions
Jul 21
•
Dr Guillermo Power
5
1
YOUR AI DOESN’T DO WHAT YOU THINK IT DOES
Designer's optimism says your prompt works. Your model disagrees.
Jul 18
•
Dr Guillermo Power
10
WHEN ONE MODEL ISN’T ENOUGH (AND YOU KNOW IT)
The discipline of using a heterogeneous AI council—sharply cutting hallucinations at a 4.2x multiplier, but reaching for it only when the bolt is…
Jul 16
•
Dr Guillermo Power
2
1
1
YOU CANNOT BAN MATH, SOFTWARE, OR PHYSICS
The US government moved to block the export of Mythos and Fable 5, with the government demanding that AI company Anthropic block all possible jailbreaks…
Jul 13
•
Dr Guillermo Power
3
1
THE WARMEST CHATBOT IS THE MOST DANGEROUS ONE
The AI industry has spent years teaching chatbots to sound warm, empathetic, and human.
Jul 11
•
Dr Guillermo Power
7
This site requires JavaScript to run correctly. Please
turn on JavaScript
or unblock scripts
