Claude has about as much "soul" as a white dude listening to Hootie and the Blowfish https://simonwillison.net/2025/Dec/2/claude-soul-document/

[codinghorror@infosec.exchange]

Claude has about as much "soul" as a white dude listening to Hootie and the Blowfish https://simonwillison.net/2025/Dec/2/claude-soul-document/

[sempf@infosec.exchange]

@codinghorror WHo do you like instead? Or just no one?

[codinghorror@infosec.exchange]

@Sempf Gemini, because ChatGPT's crawler is so Hated In the Nation it can barely get to anything any more.

[sempf@infosec.exchange]

@codinghorror I like Gemini too. It really handles search better than anyone else, and while that seems small, it really isn't.

[jaykay@hachyderm.io]

@codinghorror As a long-time skeptic, who has only found value from the LLMs in the past two months... all of this is very weird.

It's like arguing about the soul or emotions of Excel vs Numbers, or Word vs WordPerfect, or a screwdriver vs a hammer. The tool does a thing that is useful or it doesn't. Why are we talking about Pachinko machines as if they're human?

Am i supposed to be asking my literal bicycle if it is an effective altruist?

[emr@mastodon.sdf.org]

@codinghorror

> a company that genuinely believes it might be building one of the most transformative and potentially dangerous technologies in human history, yet presses forward anyway. This isn't cognitive dissonance but rather a calculated bet

These are supposed to be our least deranged AI cultists, huh.

Also: I love when there's clear examples of significant chunks of training data encoded in LLMs because people have spent a lot of time arguing to me that they simply don't do that, and...

[deane@contenttechtodon.org]

@codinghorror As a white dude who really loved Hootie back in the day, I take exception to this.

[codinghorror@infosec.exchange]

@Sempf unless you want all old data and 50 yottabytes of it locally