Knowing what to do is the bottleneck
The only curve to get ahead of is me
heys
just checking in to summarise some research into the questions from last time. It is kind of a brain dump, and i did literally just want to get it out of my head and move on.
I’ve uncovered two main issues:
the near-term threat to our own agency via cognitive offloading to generative ai.
the media-term threat to our sovereignty posted by at ai agents.
My research proposal goes into this a little and if you’re curious there are readings attached.
the future feels very volatile. Not that it ever wouldn’t be i guess. Looking into the future to create the best version of it is hard because uncertainty is pretty high.
It looks like the following situations might play out
Scarce computing resources mean limited compute power, meaning it will go to who can afford it, which will be the mega corporations like google, nvidia and apple. They will conglomerate and control the hardware and continue to rent cloud access to their models via tokens. Their customers will be the organisations, governments and companies.
An underground movement for AI compute will rise alongside this, where increasingly rare and valuable consumer compute power will be pooled via distributed networks which people can contribute to and access for ai compute. This will help democratise access, and while it already exists for nerds, will be made accessible.
In any case access will be controlled by $$$ or hardware contribution.
Access will not be universal. Current generative LLM AI that people are using to code (and as glorified google searching) is run at a loss, and is barely scalable to society-wide use cases. Efficiencies in AI software might alleviate this over the near-term, but it feels likely that demand will stay ahead of supply creating have and have nots.
The have nots are not necessarily disadvantaged: It will become increasingly apparent that AI compute while useful for many rote and routine tasks, does not solve the problem of knowing WHAT to do and WHY. AI will empower people to do whatever they want, and it might assist them in decision making, but it will never replace the intuition and
That said, people will become increasingly good at discerning what is slop. While ai threatens some cognitive powers, we will develop others: heightened sensitivity to intent and purpose. We’ll increasingly value succinct, clear and meaningful communication. It could be that everyone learns this in different ways.
What it means to create and consume will change, as ai generated/supported content continues to flood our feeds. What we can see and hear will no longer impress us and meaning will cut through noise.
AI will certainly affect the white collar job market - jobs that have no real creative input will be outsourced to AI. I think this is a great thing for civilisation as a whole - even for the people who are displaced if it’s managed well.
That’s the near-future, let’s say 10 years. AGI might come about during that time, although I think it would have to be based on entirely different computing and software paradigms than current generative models. It would certainly not be universally available and the it’s pretty hard to even guess at the future beyond this let alone prepare for it. The same basic issues will exist and be multiplied I guess.
Ultimately useful as AI output might be, it was not earned through error, therefore whether it is truth or not is almost irrelevant when it comes to human understanding. Ideally, no matter the level of intelligence, we will use it to fulfil the most meaningless and basic of human needs. In the long term, hopefully we will all fully engage in higher order pursuits being unafraid to fail, with less mortal fear, better societal safety nets. AI will support our efforts while we remain in control, like a person in a mech suit.
How does all that relate to my life?
I was considering applying to postgrad research to help map and create the future, but I won’t be studying this as I’ve seen that it’s already being done, and the opportunities around my city feel a little restrictive. Here’s the proposal (again) I was working on.
I’ve decided to keep using AI to code my projects, but not to worry about losing access. As mentioned, knowing what to do is the real bottleneck. AI will just help get things done faster. I’ll use it where i can/need and not let it control me. I’ll dabble where it suits me and where i can without investing too heavily on it. I don’t think it’s the “curve to get ahead of” that people are saying it is, i think the curve to get ahead of is my intentions and how i back them up.
I’m continuing to feel out my future and have decided to give myself a little leeway while I follow my joy (while of course constantly asking how this can be effectively monetised).
Life update: met someone who likes me for who i am and doesn’t want me to change. surreal. They’re smart, hardworking and fun to be around as well as considerate and supportive, I really don’t know what the fuck is going on.