Comment by properbrew
9 hours ago
> It seems like LLMs are great for the kind of generic work you'd usually pawn off to a new associate or paralegal
This seems to be the case with most industries that LLMs are impacting, coding, data input, legal etc.
Whilst it's great to get the time saving from not delegating to a junior (LLMs are much faster at the basic level 1 work) there's going to be this big knowledge gap that you really only gain from actually doing the mundane boring shit.
A lot of this _should_ be automated away, but we really need to think about how the next group of people are going to learn the skills and nuance to applying them. I don't really have an answer to that one other than still giving that work to junior staff, but they will then just utilise an LLM to help complete the job, and rightfully so, I'd have done exactly the same thing.
I think OpenAI tried doing this with some early chatgpt settings with "learning mode" or whatever that is. But an LLM just telling you what to do will never teach someone, there was something about having to go out and hunt for the answer through the sea of crap responses or wrong SO questions that really taught you something. How do you replicate that learning with an LLM and how do you apply that to different industries like Law?
The American manufacturing industry is scrambling to transfer some of the industrial trade and process knowledge from the retiring remnants of pre-offshoring industrial America into the two generations of people that didn’t have apprenticeships, pensions, (often) union membership with good benefits, etc. Trade schools were very expensive for industrial trade skills that were seen as dead-end careers compared to white collar jobs or non-manufacturing trades, and almost none of the companies saw any reason to spend all that money training people into expensive skilled professionals when they had all the professionals they could possibly need for decades after the industry largely collapsed, here. Keeping the workforce viable was always someone else’s problem— the individuals, the governments, larger companies— until it was suddenly everyone’s problem. They spend decades myopically chasing quarterly results before they realized they were running towards a cliff.
Unless your company had for some reason enough margin to just spend a lot more money than the competition to train and retain workers that you largely didn't need for a decade you were forced by the market to cut that corner. Market capitalism is inherently unsuited for long-term thinking. Even if everybody saw the cliff a decade ago, they all had to run towards it or be outcompeted by those who did. See also all other problems that can be externalized, e.g. pollution.
Tackling long-term problems and externalities must be done by someone outside the market.
> but we really need to think about how the next group of people are going to learn the skills and nuance to applying them
That's what school is for. For the last 150 years we've automated a bunch of things, and then we send people to school to learn how those automations work, so they can make better automations.
Maybe law school will have to be a year longer. Or maybe some forward thinking firms will have new associates check the work of the LLM to gain that experience.
> That's what school is for.
It's not.
We have advanced CNC mills and lathes, and apprentices still learn doing things manually to understand how the tools work, their failure modes, and more.
There are machines that mix drinks, but you still learn to bartend by doing it manually and making mistakes, not starting out with machines.
Same with writing. Same with software. Doing the "simple", automatable thing is how you learn how the system works.
School can't be the same. Plenty of people have tried, but being effective in the real world is different.
In theory, theory and practice are the same. In practice, they're not.
> There are machines that mix drinks, but you still learn to bartend by doing it manually and making mistakes, not starting out with machines.
You could say the same about grilling burgers. And yet, there are millions of burger cooks around the world grilling hundreds of millions of patties everyday with zero knowledge of how to grill burgers, at McDonald's and other bottom of the barrel fast food chains.
If not for the social aspect of being asked what you want for a drink and then watching it being made, bartenders would've already been replaced by mixer kiosks by now.
2 replies →
This is such beautiful rhetoric.
In the future, I think AI lawyers will be arguing with AI lawyers for the most part.
Like how people use AI to write their resume and then the HR department uses AI to screen the resumes, the 2 AI's duke it out.
Governments want to use machines to decide who to kill. As the Overton window shifts it seems natural that machines will adjudicate, too.
2 replies →
I get the impression, that's all "too abstract" for people.
Will they understand the point when football gets taken over by AI, robot players and all?
Or won't they even stop at AI pastors and pray for the second coming of "Super J AI"?
1 reply →
Still, school isn't quite the same as learning skills through work experience, is it?
> Maybe law school will have to be a year longer.
Law school is already 3 years long in the US (after 4 years of undergrad), and most agree the second two are a complete waste of time. Other jurisdictions are different, but the US market is by far the largest.
Who is "most"? Is this the same "most" that thinks college degrees are pointless and you should just go found a company (i.e. those that aren't paying attention and just want a credential)?
2 replies →
True, but have you seen the state of education in the year of our Ai-ssiah, 2026?
This is kind of a new complaint. Places like adp have being trying to automate away mundane hr tasks. Oracle and SAP have been trying to automate away the mundane admin tasks. Siemens does it for OT. There's a huge long tail of trying to solve this as if it were a problem in almost every field. This has been going on for decades and no one complains that we will lose expertise in these fields if the drudge work goes away. But bring in an AI... and everyone is suddenly concerned about corporate drudgery.
Anyone who does not do the grunt work of learning how things work at a granular level will run into trouble when their agent suggests "let's re-write the firmware!" and the user says "sure, whatever that means", the model will get stuck or break something, and there will be no one around to understand what is going on.
With strict version history that may be okay, but hopefully their agent already thought of that for them too!
Whenever you "automate away" things that are actually a part of you, you lose a part of yourself by definition.
When you automate physical movement, you atrophy. That implies ill health and helplessness/lack of resilience should your automata fail.
When you automate intellectual movement, the consequences are even more dire.
Most people lead a very immobile life already, in both aspects. They don't realize the consequences, if everybody did the same.
I think the new batch of engineers are going to think faster on their feet, but less precise.
We humans are _amazing_ pattern matching machines and learning “if you do this thing this other thing occurs” is literally what our brains were built for.
It’s just the current few thousand years of technical progress that have been a departure from this reality - we found out about analytical thinking and that by deeply analysing something you can do a shit ton more.
If you look at scientific progress before Industrial Revolution it’s all trial and error between patterns. And I feel like the new crop of devs are silently settling in into this very human and familiar pattern.
On the one hand I think it will probably be good for the “mental health” of people in this profession - you become more like an alchemist acquiring bags of tricks on what works, sharing incantations with others, because sitting down and learning _why_ it works in detail would be a time luxury few would be able to afford.
But that would mean the people under the hood that build the LLMs and who debug the really broken system would be charging the premium.
I honestly don’t think that in and off itself would be such a bad thing.
What I personally am really scared about is the price for labour that is sharply declining. I don’t worry about ASI too much, but when you read history it is absolutely full of stories of great human suffering because “labour was cheap” no need to improve working conditions because there is always someone else willing to take it up.
We had the Industrial Revolution _because_ of the plague making labour expensive and driving the benefit for innovation. Before that nobody really cared. With LLMs doing things becomes vastly cheaper, so the labour itself becomes cheap with it.
Now looking at the state of the engineered world - how absolutely shitty everything is (as opposed to what it technically could be) - there is a shit ton of stuff to be done so the work should be there to fill it. But maybe the capital will move in another direction - nobody can predict the future, I just hope we don’t end up in a dystopian hellscape for the sake of my children.
I don't think toiling away at mundane tasks is the only way to build expertise in a field, or even a particularly good one. Today's farmers barely touch hand tools anymore and they're able to produce crops just fine.
I've been having an identity crisis since generative AI started that is basically, "how am I going to learn anything with my ADHD brain when AI can do 10% of it immediately". That initial 10% would take me weeks or months to achieve on my own, and AI can do it right now, and that makes it even harder for me to bother. I hate it. I badly wish to do it myself because then I could achieve more than 10%, but now AI is right there and I just can't seem to justify the effort...