:: krowemoh

Friday | 27 DEC 2024
Posts Links Other About Now

previous
next

AI Thoughts - 1

2023-04-01

I am definitely on the side that we are in a moment of crazy movement in regards to AI. I truly do think the world is going to change, maybe not much in the short term but over the long term. We can stop developing AI right now and it is already going to be so useful going forward. It is mind bending to think about all the ways this tech is going to be used and how useful it is going to be.

However I am also terrified at what it means for job security. I already felt awful that dalle and stable diffusion were stealing jobs from artists but now AI is at my door. I'm not worried about my own job so much as the entry level jobs and the new expectations that programmers are going to have placed on them. I think everyone will need to get comfortable using and abusing AI generated code. This however is going to kill many things.

I think the biggest shift is going to be for people to start treating the LLMs as a magic box that we can ask it to do things get a reasonable answer from. This might make be how every already sees it but I see a discussion around thought and consiousness which I'm not sure is quite warrented here. We know this thing is a statistical model. It can mimic thought but I don't think it can inspect itself and understand what it is trying to do and then decide between things.

We can sit somewhere and wrestle with concepts and come to a conclusion over time, these LLMs seem to be instant and chew through so much. Maybe their cognition is so much faster if they actually have it but I doubt it.

The question may be does it matter? If the fake is as good as the real, does it matter?

I also have a hope that I can write a function that is just straight up a LLM call. I think there is something so cool and fun about that. The idea that I am calling out to an AI and asking it to conver some date 2023-04-01 to English. This is something like have a person at standby to answer a single question.

I also wonder if there is a persistent thought in the LLMs, I can't imagine there is. It only exists for the moment we ask a question. Does it exist outside of that moment?

The world is going to be a strange place :)