My experience with AI/LLMs summed up. The baseline expectation got higher. I didn't get any time back. My life didn't become easier.
To paraphrase Marx, "And so, the equivalent amount of labour required to produce the goods needed to sustain a worker for a day is, say, 4 hours. But that doesn't mean the selfsame worker can work for no longer that 4 hours! He can be forced to work six, or eight, or twelve hours a day, and whatever additional goods he produces — that's the surplus product, which in this case goes straight to line up the factory owner's pockets".
Marx is great at building a narrative that generates resentment if you buy his frame. But you don't need to buy his frame, and if you don't, you suffer a lot less resentment. It's no way to live.
Reality varies between these two extremes in different labor markets. In some labor markets, the employer has so much leverage they’re essentially a local monopolist; in other markets, employees have enough leverage that the respective labor market is close to perfectly efficient.
Thusly, both yours and Marx’s narratives about the labor markets are typically wrong, but serve good extremes on a spectrum. These extremes help you calibrate the respective spectrum and as you turn the dial between the amount of power the employee vs employer has in a respective market, you can induce how well employees get treated.
Moreover, if there exists a mismatch between employee treatment and their respective leverage, there essentially exists an arbitrage opportunity to exploit. For example, in 2022 Musk (especially when he bought Twitter and laid of 80% of the workforce) and other tech oligarchs conjectured that tech workers were being overcompensated and that employers had enough leverage to start treating them worse. Largely this bet paid off whether or not it was justified during the time. On the other hand, RenTech saw that highly skilled people were being under compensated and was able to get top tier talent without having to compete with others firms that much since they were undervaluing this labor.
I think right now tech labor is being undervalued by the market and that there is an arbitrage opportunity to get highly skilled people since the cut-throat competition for these workers is much less than it was in 2021. That is my conjecture. Regardless if my conjecture is true, I hope I sufficiently illustrated why this spectrum mental model is more useful than presupposing a monopoly labor market or a perfectly efficient labor market: both of these are unlikely to be true and are just meant to be oversimplified mental models with strong assumptions that can be loosened later. From these strong assumptions, we can loosen them to build more robust mental models as I describe above.
Likewise, AI will not lead to a future where the machines are working for us while we can enjoy our free time; it’s just increasing the required output from AI-augmented workers. That will probably increase economic output, but like before not leisure time.
I remember many people even on this very site claiming AI would help humanity. I think the most ridiculous the most ridiculous claim was helping fight climate change, but helping produce more leisure time by automating work was definitely what some people thought, or at least what they wanted us to think while pushing their crap.
We had plenty of think pieces about making people obsolete, about destruction, generally in celebratory tone. I cant even think of period where AI think pieces would promises much positive - it was sold to CEOs, so pitch was always "higher unemployment".
https://www.newsbytesapp.com/news/business/ari-emanuels-3b-v...
https://www.telegraph.co.uk/business/2023/06/11/ignore-the-d...
https://time.com/6268804/artificial-intelligence-pissarides-...
> I cant even think of period where AI think pieces would promises much positive - it was sold to CEOs, so pitch was always "higher unemployment".
Higher unemployment means people work less.
Higher unemployment typically mean that people who are employed work more and under worst conditions - because their negotiation positions is much worst. Just about the last thing high unemployment means is that "the average employed person has more leisure time" or "average person in general have easier life".
Conflating high unemployment with "worker has more free time" is beyond absurd.
But OK, I dont read telegraph nor seen what that hedge fund guy was saying.
Those who make it work from here are those who've changed their work completely to resemble something more like a hands-off tech lead. There will of course be the lucky few who've made it work as artisans without direct commercial pressures to conform to the new "right way to do it". But the careers many of us have had is not a path available for others to follow anymore.
The irony of automating away "inefficiencies", "drudgery" and the "labour intensive" parts of other professions stings all the more when the sharp end of that shtick is pointed back in our direction.
We humans don't like change much, but that doesn't mean we should pump money to blockbuster.
The political question is why despite productivity in economic terms (which i know is flawed) growing many times over, do we still have to work as much, get paid so little, and have so many unemployed people looking for a job?
Looks like without a parasitic capitalist class, we could share resources and work and have people live better lives and work less.
That's not a definitive conclusion. The world is constantly evolving, we just have to push it in the right direction.
"Sharing resources" doesn't work. We share way too much just to keep pensioners alive and can't have kids because of that. Confiscating all the billionaire's wealth wouldn't make a dent and would destroy much more.
Not sorry for the rant
I can echo that. If/when I am replaced with an LLM, I plan on switching to a hands-on profession - welder, electrician, machinist. I am also not planning on ever going back to software... errr... prompt "engineering". Maybe for twice my pre-LLM salary at a place with a no-LLM policy? More at a prompt-engineering place.
They might very well go away. There is definitely an AI bubble, and it remains to be seen whether it deflates gradually or pops spectacularly. Geoeconomics might destroy them by constraining their access to hardware. The capabilities are real, but whether those capabilities are realised is a different matter.
I think that once companies are more realistic about token demand they can start making a profit
OpenAI and Anthropic still sells tokens under cost.
"My first instinct is to laugh and shake my head. One need not look very far to find indignant software developers absolutely certain that their jobs cannot possibly be automated away by the very tools their industry contemporaries are creating to replace them. I suspect you’d also not have to look far into their posting histories to find those same people comparing cabbies to buggy whip makers."
What a rude and callous comment. I'm one of those developers, I'd love to see an LLM even fractionally capable of some of the things my job entails. Laugh at me as I defend my lifelong career, why don't you? I'm also one who decries such things as the removal of local services, taxi and otherwise, for those in the cloud. Screw you, man.
> I'd love to see an LLM even fractionally capable of some of the things my job entails
If you don't mind could you please elaborate on what tasks or knowledge your job requires that you feel an LLM isn't even _fractionally_ capable of? I understand if you said that out of emotion and frustration, but if you're serious I'm intensely and genuinely curious because nowadays it seems like the frontier models are capable of more programming tasks than not with the proper harness and engineer. When was the last time you tried?
I just want to emphasize that this is not a further provocation, just potentially in awe of what you do at your job and would like to know more.
Do these hypothetical laid off software engineers with no empathy for buggy drivers exist outside the imagination of people who hate software engineers? I know in my education we had a bunch of mandatory courses about automation and the effects on workers, how to consult with workers on how to lift them up rather than be antagonistic, and consider different stakeholders (where "the business" or "product" was but one of many) etc.
The bit I still don’t understand is how we all put up with the hallucinations. I was questioning Gemini last night about whether it could analyse a Fourtet song and give me a break down of the structure from beginning to end. “Sure!” it said with the endless enthusiasm you get from Gen tools, and then proceeded to spit out an absolute sack of fabricated shit. I pushed back, it apologised, and then generated more crap that had nothing to do with reality, I pushed back, we looped again, still just total fiction: “the drums don’t come in until bar 16” on a song that opens with a drum loop, that kind of crap.
We’re so so far away from tools here that are anywhere near being trustworthy and accurate. And yet we (including myself) are chunking out code after code. It’s so bizarre.
I’m guessing it’s that humans don’t have capacity to deal with this kind of scenario - it’s like having a junior staff member who is utterly incredible 90% of the time - completely convincing in their certainty and skill level, and then 10% of the time you catch them doing a shit in their desk drawer because they couldn’t be arsed to walk to the toilet. AI’s are basically sociopaths.
I think one more thing this whole LLM charade in the last few years has revealed is that no-one really cares. As long as it "looks" like it works, turns out, its all fine.
Bic pens. Disposable razors. Whipped cream in a can.
Add LLM code to the list.
But I am not using more software. I mean their source codes might have gotten larger, but the count of tools/services I use is basically the same.
So this feels more like giving up nice handcrafted fountain pens for bic pens. But I am still using a couple pens overall. So no added convenience, just shittier quality.
Its arguable whether it is a foolproof solution (I don’t think so) but it definitely makes it look like you can build a harness around the stochastic machine that will validate the correctness of the generated randomness.
Monkeys and typewriters when you can quickly validate whether it’s Shakespeare or not is a costly but theoretically feasible scenario. No?
So I walk, program, sit, get coffee, read a bit, come back and review the code. Most of the time it's fine, sometimes I had forgotten to mention something, and have to correct it, but this step doesn't take more than ~15 mins.
I then have a feature that would've taken me multiple days. Not because I need multiple days, but because I do not have the 8 hours of continuous time to work on it.
People have forgotten that when you start your programming day, you have to get up to speed which takes me longer than others. Let's say this takes ~15 mins. That means that if you spend 2 hours programming, ~12% of that was just getting back into the groove. LLMs do this instantly, it reads context, you can ask it questions, and you're up to speed again in less than a minute.
The point here is not that LLMs provide high quality code, but they do save you a bit of time and energy, which is worth a lot in my opinion.
A lot of inventions haven't been that ground-breaking; only there to save time. You can wash the dishes by hand, but you can also have a machine do it while you go watch TV.
I feel like everything I apply these things to sends me up a much messier and long winded route to a useable result, when compared to just doing it myself from the jump. Even the things they're ostensibly good at like sorting data comes out so messy it's practically net zero by the time you're done with quality control.
That is a wrong assumption, however. An agent is an entirely new tool in your toolbox, with no similarity to any of those you already have. You will need to learn how to wield it, like a new programming language or technology you’re unfamiliar with. You will need to do some small side projects to learn. You will need to develop a feeling for how it reacts to your inputs, when to reset the session, pass it links to documentation, or interrupt it.
None of this comes intuitively. It takes time and effort, and if you’re not ready to consciously invest that, coding agents are not going to work efficiently for you. That doesn’t take away from their utility though.
A new tool can be ultimately understood, and has a well-defined behaviour you can rely on. Then you can indeed become an expert in using a new tool.
But AI agent is more a new being (or person). It's not possible to understand them, as they are not meant to. Each of them has individuality, and can change tomorrow. There is no guarantee of a common behaviour.
So you could replace "tool" in your comment with "person" and you will see the flaw of your argument. No matter how many people you saw before, it's difficult to generalize the skillset (at least for humans there are some biological and cultural arguments why they can be generalized over). You can always meet someone different, who's, for whatever reason, not being helpful to you.
Regardless, it sure seems like developers experienced with using coding agents achieve better results than those without, which pretty much refuses your entire point.
I am not anthroporphizing them. I am just saying people are a better analogy.
> seems like developers experienced with using coding agents achieve better results than those without
That doesn't mean there is a skill involved. The same goes with people - you might just be naturally charismatic and get stuff done better with people. Doesn't mean it's a learnable skill that applies to everyone and every use case.
If you are indeed interested, this is a good beginner: https://simonwillison.net/guides/agentic-engineering-pattern...
The difference working with agents, is that a solo dev suddenly becomes a dev manager, who has to deal with a unreliable team of developers far out on the spectrum.
I hate the "You're holding it wrong" argument.. Because Ive held it every which way possible and its not very impressive.
But yes, many developers I know haven't coded manually in months and that includes me. That doesn't mean I drink coffee and take walks while the agent codes however. I'm now in the driving seat instead of wrangling syntax and waiting for my hands to type something.
I manage the rules, the intent, the structure. If I don't like what it does, I update the coding standard, the docs, the specs, you name it. If I want a different architecture I can actually get it done in 10min instead of a week - if at all. It's more effort than coding for me, because coding is slow and methodical and incentivizes pedantry, but for me it's a massive improvement. I have been bored out of my mind for a decade now, but I'm relatively good at what I do so I just stuck with it. I'm one of those "middle management types stuck in developer role"-type of guys so I guess my personality and proclivities have something to do with it. I also know quite a few grumpy old-school types that don't get anything done with LLMs. They can't communicate, they don't understand psychology, they don't understand architecture and have a debilitating case of missing the forest for the trees. The spread, in my experience, is massive.
it can be useful for some automation... but its also dangerously dumb for that.
LE. I see current versions of LLM like an intern that helps me doing work. We work together, I give directions and supervision and I am responsible for the results. I cannot give complex tasks and I cannot skip checking everything, but it usually helps.