Dare Obasanjo3d ago

Every tech person I know who has gone deep into using AI tools comes away with the same impression, with the right context Claude can already do a surprising chunk of my job and this is the worst the tech will ever be.

The question is mainly how far behind big tech the rest of the world is? Optimistically it’s 3-5 years away given adoption curves but I’ve been surprised how quickly things have changed in big tech since November.

https://sahajgarg.github.io/blog/cognitive-labor/

Show threadKevin Miller
@carnage4life I think production of software for AI is “easy” because we have a huge public corpus. How many other industries have a digital double of their knowledge that extensive?
Mar 20 at 2:57amWeb
Show threadDare Obasanjo3d ago

@kevm I think a key question is how unique is the software industry?

It’s clear that creating software code, videos and images also had the benefit of a large corpus of public data. What else fits the bill?

Show threadKevin Miller3d ago

@carnage4life I remember thinking when music was being ripped and traded that any digital media would become a liability in a way to creators.

I never dreamed coders would do a similar thing to our future selves via open source code.

Show threadアイアンハンドガンダム3d ago
@kevm @carnage4life are open source models the answer going forward? Can the community create a model that is good enough and that can run on local hardware?
Show threadKevin Miller3d ago
@mrt181 @carnage4life I’m a big fan of open LLM models. Though, my point was that the extensive open source codebase was the perfect training corpus that coders created to make our world better. Ironically in the end it removed the act of coding from their jobs.