Henrik Carlsson's Blog

All things me.

posted this note on and tagged it with AI

There are plenty of things that can be improved by LLM’s and similar technology. It has been done plenty of times prior to the current A.I. bubble. But this kind of slop, posing as answers, needs to disappear. What is the point of this, other than virtue signaling that Google is an ”A.I. First Company”, aka drinkers of the Kool-Aid?

My query to Google was what word count I should equate to a page of text. The answer claims that it depends on a lot of factors and gives a rough estimate of 250-500 words. That is probably reasonable but it then goes of to claim that the bigger my line-spacing, the more words I can fit on a page. It’s just so hilariously dumb. Given how obvious it is that at least parts of this answer is wrong, I had to look further in the actual search-results. No time was saved for me, no value was created, nothing was gained.

This also ties back nicely to this.

posted this note on and tagged it with AI

I’m skeptical about OpenAI in general, because while I think AI is so powerful that aspects of it will legitimately change the world, I also think it has been overhyped more than just about anything I’ve seen in my three decades of writing about technology. Sam Altman strikes me as being a drinker of his own Kool-Aid, but it’s also his job to make everyone in the world think that AI is inevitable and amazing and that his company is the unassailable leader while it’s bleeding cash.

Jason Snell: Sam and Jony and skepticism

This paragraph from Jason Snell perfectly encapsulates my feeling about OpenAI, and so many other ”AI Companies”.

Replies and comments

posted this note on and tagged it with AI

It’s getting increasingly hard to use Office365 without being forced to interact with Copilot. If Microsoft love AI, good for them I guess, but I just want to read an e-mail or create a form or something similar. Don’t put your prompt in the way of me getting my job done.

posted this note on and tagged it with AI

[LLMs are] not search engines, and it doesn’t matter how much Google tries to force AI into its search, it’s not going to deliver what we need. LLMs are NOT answer machines. They’re guessing machines. And any guess has the potential to be wrong.

AI in search is a trap: it’ll ruin trust

Spot on!

posted this note on and tagged it with AI

“[T]he internet wasn’t built to be a factory for engagement metrics and AI-generated content farms. It was built to connect us, not silo us to pad a wealth-extremist’s bank account.”

The Fediverse Isn’t the Future. It’s the Present We’ve Been Denied.