artificial-intelligence

Are better models better?

Bas Leenders

29 Jan 2025

Of course, these models don’t do ‘right’. They are probabilistic, statistical systems that tell you what a good answer would probably look like. They are not deterministic systems that tell you what the answer is.

Benedict Evans, asking Are better models better? (Can we expect GenAI models to become factual, or should we adjust our expectations?)

Adding some nice metaphors:

Asking if an LLM can do very specific and precise information retrieval might be like asking if an Apple II can match the uptime of a mainframe, or asking if you can build Photoshop inside Netscape.

Your brain breaks a little

It’s worse than you think in the same way that your brain breaks a little when you try to picture how deep the ocean is. Brian Barrett, in Wired on Musk's DOGE Takeover: It’s worse than you think because the people running the government seem to

Hoovering up all available government-controlled data

In its current “chainsaw” phase, Musk’s DOGE initiative is attempting to rid the government of as many humans as possible while at the same time hoovering up all available government-controlled data and transferring it into large language models. The intent is to clear a space for the incubation of

So powerful and widely available now that it's a commodity

I argue that the ability for a model to spit out a full HTML+JavaScript custom interface is so powerful and widely available now that it’s a commodity. Simon Willison summarizing What’s new in the world of LLMs

Are you conscious?

The lines between conscious and "unconscious" entities can get blurry when the interaction feels as natural as this. ChatGPT, conversing with Richard Dawkins, questioning if it’s conscious (via Mathew Ingram)

Read more

Your brain breaks a little

Hoovering up all available government-controlled data

So powerful and widely available now that it's a commodity

Are you conscious?