Skip to content

Work

Your LLM Doesn’t Write Correct Code. It Writes Plausible Code.

So beware. “LLMs optimize for plausibility over correctness. In this case, plausible is about 20,000 times slower than correct.” https://x.com/KatanaLarp/status/2029928471632224486#:~:text=LLMs%20optimize%20for%20plausibility%20over%20correctness.%20In%20this%20case%2C%20plausible%20is%20about%2020%2C000%20times%20slower%20than%20correct.

Never bet against x86

Standardization matters. “The problem with Arm processors in the desktop (and laptop) space certainly isn’t one of performance – as this latest design by Arm once again shows. No, the real problem is a complete and utter lack of standardisation,… Read More »Never bet against x86

The Uncanny Valley

I got this feeling every time I see a document at work “polished” by AI. “First, there’s a corpse alert. Corpses are dangerous, and something that’s alive/not alive is a warning sign. Same thing with zombies.Second, imposter alert. Imposters are… Read More »The Uncanny Valley

Outsourcing vs. AI

I remember a few years ago a company has fired a guy, because he has outsourced his development job to India and China and just coordinated the work collecting his salary and paying the team he managed. Nowadays management basically… Read More »Outsourcing vs. AI

Computer Says No

Little Britain AI. “If AI takes over the coding, that layer of knowledge disappears. Not just for the user, but for the software engineers themselves. The product becomes a black box. And yes, perhaps new craftsmanship emerges in directing the… Read More »Computer Says No

Team Meet vs 1:1

I often struggle with meetings with multiple of my direct reports. What should I discuss there vs. 1:1s? It can save me time to update all at once on some topics vs. discussing 1:1, but then it becomes less deep… Read More »Team Meet vs 1:1