first read as "bearables"

I'm really excited that this thing from Ive could be a reality because it's a perfect package to confirm so much of my theory around what makes a bad [tech] product

Real problem? The product exists because it's possible. Not because it is the result of a deliberative design process to find the best way to satisfy a concrete purpose.

Every AI wearable has the same challenge - make a hardware container for chatGPT that can't be replaced by a smartphone app.

The most obvious way to do that? Fashion.

The problem? Fashion works by being visible.

From portable tech that you didn't want to hide because there was no question about its purpose.

The blurrier the purpose gets the more the products need to be hidden.

Deadbert

in the arms of phil

useful life is absolutely not indefinite

> models can update their knowledge via cheap search queries

i'm not clear what you mean by this, cos to my knowledge this isn't a thing

fine tuning and RAG can address this a bit, though at best you can add new things and not remove deprecated things, so expect coding models to decay

"80–90% of computing power for AI is used for inference" www.technologyreview.com/2025/05/20/1...

this is fuzzy as hell of course, but inference = answering queries

what numbers we have show inference is currently 80-90% of costs?

these are all fuzzy numbers of course

and if they can't train, the models go out of date

(after dumping two hundred gallons of e-waste into a black neighborhoods drinking water) "maybe you dislike this because you're neurotypical"

that's the opposite of what I said. at this point you're deliberately wasting readers' time/

Friend of mine who paid £1500 for a chimp still tells me it will come good. Market value of about 8p currently

show us your measurements. without these, your anecdotes are literally useless and a waste of everyone's time.