Alessandro ReMar 2
Gerben Wierda
One of the best essays on the GenAI hype I have read so far: https://www.baldurbjarnason.com/2025/trusting-your-own-judgement-on-ai/ by @baldur
Trusting your own judgement on ‘AI’ is a huge risk

Writing at the end of the world, from Hveragerði, Iceland

Mar 2 at 9:42amWeb
Show threadJohn Faithfull 🌍🇪🇺🏴󠁧󠁢󠁳󠁣󠁴󠁿🧡✊🏻✊🏿Mar 2
@gctwnl @baldur Excellent! I missed this earlier -thanks so much 👏🙏
Show threadSebastian LasseMar 2

@gctwnl @baldur

Funny, came in just after reading the story about the woman who decides at Antrophic what is "a good character" –
https://www.spiegel.de/netzwelt/kuenstliche-intelligenz-wie-die-ki-bei-anthropic-einen-guten-charakter-bekommen-soll-a-eacaa011-8063-48d2-8818-40cb6f78393b
-->
wondered: Why would you call yourself a philosopher if you want to be God.

Heikle Phase der Entwicklung: Diese Frau entscheidet, ob KI einen guten Charakter bekommt

Die Philosophin Amanda Askell soll bei der Softwarefirma Anthropic dafür sorgen, dass sich KI-Modelle an Regeln halten. Sie sagt: Menschen seien dafür kein gutes Vorbild.

DER SPIEGEL
Show threadBrantley HarrisMar 2

@gctwnl @baldur I appreciate the central message that LLMs short circuit the brain and we can’t trust our conclusions of it. I want more exploration of that. They are smoke and mirrors and flattery. They produce quick often inch-deep results.

However the crux of the argument seems extraordinarily reductionist. If there is no science and I am not to trust my own experience I will forever be locked in paralysis except for *maybe* highly tested medicine.

Show threadBrantley HarrisMar 2

If AI analysis is corrupt simply by being in the current landscape, then this analysis is also corrupt and all analysis is, and so we are painted into a corner of information paralysis.

And so then this becomes an easy reinforcement of bias for those who have already decided to avoid AI.

And if all analysis is corrupt, then in fact, Tim Bray’s analysis is the *most scientific*, since he is at least citing a survey of many examples, even if they carry the corruption.

Show threadneoludditeMar 2

@deadwisdom Except that "smart developer tries LLMs, thinks they're good" is the same example repeated many times.

And yes, you're correct: a lot of analysis performed by non-experts is suspect if not outright wrong. There are entire branches of maths devoted to picking out all of the ways analysis can be suspect.

Including this blog post, yes. Do I agree with it because it's right? Or because it reinforces my beliefs? Hard to say, but it certainly gives some good counter arguments.

Like Feynman said, "The first principle is that you must not fool yourself and you are the easiest person to fool."