An intersting analyzis of the rise of the personal computer and the dotcom bubble, and how they current AI bubble might shake out.
OpenAI Has a Fix For Hallucinations, But You Really Won’t Like It
This explains why LLMs are gaslighting or lying to you all the time. The most worrying part is that the benchmarks for LLMs rank the answer “I dont’t know” at the same level as a wrong answer. This creates the current models that are rewarded for producing as high answer rate as possible, at the cost of lying sometimes. How many people would stop using ChatGPT if it answered I dont know to 30% of the answers, a conservative estimate based on the research.
Loved this article. I’ve always thought most self-improvement tips like “make your bed every morning” are dumb. If I had the mental capacity to make my bed when I got up, I would. It’s a symptom, not a cure.