I read these articles about how much of a suckers bet AI is and start feeling existential dread about the coming market crash.
Then I remember I just did a weeks worth of work in a few hours and feel a lot better about my prospects at least.
AI won't replace people, but it will make people who know how to use it vastly more efficient. In the same way that a tractor made a farmer more efficient.
I am still a bit so-so on stuff like Codex and Cursor, but ChatGPT has undeniably helped me get more work done, if for no other reason than unblocking me with stuff.
I’ve mentioned this before, but it’s relevant. Most of my adult life has followed a pattern: I work on a personal project until I get blocked on some bullshit detail (e.g. some arcane Linux server error), get annoyed because I am not making money and the project is no longer fun, and give up on the project about halfway through.
Being able to paste server logs and weird error messages and having ChatGPT generate recovery scripts has been extremely valuable to me. I am not even sure it’s necessarily reducing my workload, but it has absolutely allowed me to focus much more interesting and fun parts, and I finish a much larger percentage of my projects.
I am probably not learning as much as I would if I had powered through the issue, but I am definitely learning more than if I had abandoned it.
I agree until the end - won’t replace people - it will replace people. There are tons of people who a) can’t or won’t learn how to maximize their value with the fools, b) some jobs will outright be automated and they’re jobs that require little executive intelligence and judgement. Not everyone is smart - the median IQ is 100. I expect a hollowing out of non manual labor work under 115 IQ. Further, world models and robotics will start hollowing all but the most poorly paid manual labor or the highest skill, which most people are incapable of. There will be a large hollowing out of labor across the low end.
The fact we haven’t fully automated everything in ~3 years seems to be the proof this isn’t really happening. It takes at least a decade to iron out how a new technology fits into things how to exploit it efficiently, and this specific technology requires a lot of subtle research and building tooling and infrastructural code, processes, and techniques we’ve never built before. It also has a large physical infrastructure requirement.
However for those manual tasks above (effectively any sort of coding, classification, labeling, task oriented data work) my experience is LLMs have much higher precision and recall at any scale of work without labor issues around hiring and ramp up/idle time, let alone cost. I’m not saying this is good - but it is absolutely true and in my last two jobs we’ve eliminated massive numbers of roles for people and we are just beginning in the cycle and not very mature. By the end I expect as many as 20k+ reduced, maybe much much more, from my current company in these sorts of roles.
Did you pay for that AI to work for you? How much would you be willing to pay for it?
Tractors undoubtedly increase farmer efficiency, the evidence is clear to see, even when accounting for all costs necessary to design and produce tractors. There’s even room for farmers and tractor manufacturers to generate economic profit.
> Tractors undoubtedly increase farmer efficiency, the evidence is clear to see, even when accounting for all costs necessary to design and produce tractors. There’s even room for farmers and tractor manufacturers to generate economic profit.
Aren’t small independent farms struggling because of debt loads are really small returns?
Is this what AI will do as well? Enable large players consolidate while anything bigger than hobby work becomes very difficult and expensive for individuals.
Sample size of one, but I pay $20 a month for ChatGPT and I feel I get much more than that of value from it.
I would honestly probably be willing to pay $75-100 for ChatGPT if it came down to it. I feel it makes some tedious jobs a lot less horrible and in turn pays for itself.
The fact that LLMs are somewhat useful isn't really the subject here. The problem is that to be worth what they're evaluated at they'd need to be thousands of times more useful than they are. They can be useful, and extremely overvaluated at the same time, which still ends up in a crash. All the actors are bleeding money right now, would you spend $500 a month on an llm ? $1000 ? how many people would ?
You also don't account for the infinite flow of slop being generated 100x faster than any productive work: "Deloitte to pay money back to Albanese government after using AI in $440,000 report", https://news.ycombinator.com/item?id=45500485
How much money, resource and time is spent on generating/hosting/discussing/cleaning up this slop ? How much of social distrust is created by LLM generated sociopolitcal discussions online ? &c.
Then I remember I just did a weeks worth of work in a few hours and feel a lot better about my prospects at least.
AI won't replace people, but it will make people who know how to use it vastly more efficient. In the same way that a tractor made a farmer more efficient.