Comment by ThalesX | Hacker News Reader

Comment by ThalesX | Hacker News Reader

Parent Full thread

I'm a software and product guy; what I don't get, is how we're going to replace us lowly engineers and not a substantial percentage of a company's Human stack.

Close to engineering, what about the SCRUM masters? The testers. What about the product owners? What about devops? Further from us, what about the people signing up on our vacation? Or the ones signing up on our daily budgets when traveling, or hell, even the ones we interview with.

In my closest group of friends (we're all seniors in our domains, and very honest with eachother), I find that only the construction worker's job should be safe. And compared to myself and the devops guy, most others have what they themselves describe as trivial and bullshit jobs. Join a meeting, do some paper pushing, some document signing, a little coffee and the day is over by 1PM.

Am I seeing all these AIs replacing programming because I'm on a board where maybe a lot of us are programmers? Is it the same for other roles? Wouldn't it make more sense to have the interview process automated by LLMs if they're capable of building great software, before we replace those hired?

I'm very confused by all the hype when matched with my experience of using LLMs daily for the past years.

Well, I am not sure why this is confusing. Engineers do things the rest of the company don't understand and everyone depends on.

Removing dependencies and lack of clarity are two principles you would apply to your own code base. The Human stack, like you called it, is doing the same with the company and staff : remove dependencies and increase clarity.

I am not saying this is the correct approach, but this is the approach.

At the same time, we can understand that real product building and architecture require more than codegen, with or without AI.

Generally, I look at the AI recent progresses similarly to what spreadsheets did to accounting a long time ago. It removed a lot of the dummy work and make everything more efficient. But we still need accountant and they are not wasting their day doing basic math. Of course the one who knew only how to make fast math but not accounting or business logic didn't make the cut on the long run, but I don't think anyone would want to come back to accounting like it was 60 years ago.

AI doesn't seem like it's able to automate software development yet. It's not autonomous enough. The examples shown in the paper are toys. It might become more capable after training LLMs against code execution, to iterate based on feedback, because that way of generating training data is scalable. But strangely I haven't seen much progress yet.

Another domain where I expect to see lots of progress but nothing shows is training from chat logs. I estimate "OpenAI" is making 1T tokens per month from 100M users. That's some serious corpus of text, on-policy data, alternating LLM outputs and human feedbacks, we should see iterative improvements. Why are there no papers? weird

Forget the question of full automation. Think about the percentage increase in productivity, meaning companies can fire lots of people without impacting output. Obviously companies can also choose to keep the same number of people and simply produce more. But not all will.

It doesn't mean that companies "can fire lots of people without impacting output". It just means we're going to produce more instead of working less. Competition will raise the bar in affordability, quality and novelty. That means everyone is going to compete on a different level. Working like it's still 2020 won't cut it anymore.

That's what I said was one of the possibilities. But it is not the only possibility, and it will become less likely over time. Eventually the lift operators and steam engine dudes get deprecated. More trains are running, but the steam engine dudes got fired because they're not needed. Maybe you'll be the first non-customer facing profession in history that doesn't go through deprecation in the face of automation. Good luck with that bet.

See Appendix D in https://arxiv.org/pdf/2303.10130.pdf OpenAI’s “GPTs are GPTs: An early look at the labor market impact potential of large language models”

It’s not just construction workers. Still safe to be a pro athlete and a short order cook, too!

I feel like no matter how productive AI becomes, it will always be more productive when paired with humans. Since all companies will have AI, and since they will continue to play the same game as before--sell a good or service that provides a better value than the competition--that means that humans will still be the differentiator.

This is what Kasparov said about chess. He was wrong. The “centaur” phase lasted for a while but after a few years humans only make things worse. In fact that is the famous last words of humans in every area where computers have surpassed them.

"Productive" may have been a poor word choice. When faced with a sea of AI-driven options people will be looking for a differentiator. The human touch will be the differentiator. There will probably be a lot of efforts to authenticate human involvement.

Consider the alternative. A million models all competing for human dollars. How can any of them succeed if they are all interchangeable? It stops mattering that they are better than humans if they can't make money.

True, but at the same time humans seem to have a self delusion/bias of efficiency.

If we were really Homo economicus there wouldn't be so many bullshit jobs right now.

Homo economicus would be excited about the 2024 computer chess championships. The idea of grand masters would be a relic of the past.

Homo economicus in 2050 would be a single CEO with everything else done by AI. A real human though will hire another human for some bullshit job just to tell the owner how great they are. Over time, that human with the bullshit job will gain status in society and hire/convince the owner they need to hire human number 3 to tell human number 2 how great they are. Hell, a real human will waste money on an office and hire other humans for bullshit jobs just so their parking spot feels like something special in the morning.

Magnus Carlsen couldn't be doing better even though Homo economicus would no longer be impressed at all with a rating of 2882.

Homo economicus would think that rating sucks but humans have nothing to do with Homo economicus.

Don't really understand the trends myself (don't mean this in the positive or negative sense, just some grays popping up in my stubble), but the more people I know or meet that have jobs in IT, I have to notice that the tiniest percent of them are programmers. And it keeps getting more and more diluted.

Agree. We developers try to create AI to replace ourselves first while making the bosses rich

If you employ an AI to do software development for you then you are now a boss, congratulations. Getting rich is still up to you though.

I think we’re seeing the beginning upward trend of an exponential curve for humanoid robotics too, which will likely take 90% of construction jobs because humans are especially poorly equipped for difficult manual labor on a daily basis.

IMO humans are quite well equipped for manual labor, much more so than for most office jobs. Millions of years of evolution perfected us for eye-motor coordination, not writing code or attending Teams meetings. Building robots that can do it as well as humans is going to be a difficult task.

humans are at a disadvantage from robots in that:

1. they need to sleep 8 hours a day

2. they will refuse to work more than 9 hours a day in many cases

3. they need frequent breaks

4. they are fueled by biomatter which is massively resource consuming to generate and deliver

5. they provide inconsistent results

6. they fight with other workers

7. they can be dishohest

8. bodies break down irreparably after a couple decades of this style of work and require hundreds of thousands of dollars in medical care

9. it takes 20 years to produce another viable human to serve as a replacement worker, and there's very little you can do to influence the general supply of total workers

10. they follow instructions poorly

11. they take an incredible amount of time to learn

12. they require extreme levels of safety precautions that an all-robot work crew would not need

Reply on news.ycombinator.com