💯 I remember watching the Vows section stream in late on Saturday night.

True words.

I’ve lowkey switched to NYT trending as my homepage. www.nytimes.com/trending/

It’s still alive tho www.nytimes.com/timeswire

ChatGPT is pretty remarkable. My avatar is ChatGPT + a selfie I took + a few back-and-forth prompts. After a while you can identify its handiwork easily, though that’s true for human illustrators too.

$850 is steep. The target market can’t easily afford that. Should have aimed for $450, the price of a gaming console.

Seems like the kind of thing @etymology.substack.com.web.brid.gy might know.

Some important questions to think about: How long ago? Things have come a long way since even a month or two ago. Which model? There are dozens. What was your prompt? There are ways to ask that markedly improve the results.

There are many use cases. I think you mean “AI Mode” / generative search. But ask a ChatGPT “deep research” model for a backgrounder on a person or company and you’ll get a useful and thorough answer. You shouldn’t immediately publish it, but that goes for all first drafts.

The technology has moved on from the early hallucination problems. The latest models are worlds better than a year ago when they were making up court cases. Deep Research models in particular are astonishingly useful, and RAGs are in public-facing production all over the place.

Not sure if you’re trying to engage or just QT-dunking but reliability has improved tremendously since the early well-publicized flubs, and with good prompt engineering and a human-in-the-loop checking the output, GenAI can be both reliable and useful.