Quinn Norton wrote an insightful piece about large language models and evolved human behavior at EmptyWheel. I thought this observation was particularly on point: “They don’t even have to be engineered to charm us, and they aren’t. We’ve been engineered by evolution to be charmed.” Continue reading

In a study just published in CHI 2025, researchers found “reductions in cognitive effort and confidence” among “knowledge workers” who made use of AI. Now, this is real research and definitely worth further investigation. But also what they found was self-reported reductions in cognitive effort and confidence. In other words, they just did a survey… Continue reading

Kwaai is an organization working on personal AI. Part of that is figuring out what “personal AI” means. Looks interesting, and you can join in at various levels — the lowest level is “observer,” which is akin to subscribing to their newsletter. Not that they have a newsletter, exactly. Anyway, let’s see what can come… Continue reading

“The cost to use a given level of AI falls about 10x every 12 months, and lower prices lead to much more use. You can see this in the token cost from GPT-4 in early 2023 to GPT-4o in mid-2024, where the price per token dropped about 150x in that time period. Moore’s law changed the… Continue reading

From Jeffrey Emanuel, one of the best analyses of the advances in DeepSeek AI, wrapped in an investor’s rundown of Nvidia’s valuation, business, and future. This is long, but full of great information and insights. Here’s another good DeepSeek post, written as a FAQ by Ben Thompson. You might notice that it’s a blog entry… Continue reading

From IEEE Spectrum, an interesting analysis of how AIs make mistakes, as do humans, but because the AI mistakes differ from human mistakes, the expectations we have and the infrastructure we’ve created to recognize, reduce, and correct for human mistakes doesn’t fit AIs as well. We’re actually pretty good at dealing with mistakes — to… Continue reading