[everdrive]

I was never excited for automation. Automation doesn't mean we do less. It means that we do as much work, and now also the work has a higher complexity ceiling; you need to understand the systems that are being automated, and need to maintain the automation. More things are possible, but everything is more complex, and of course, you still need to work 40 hours a week. Products don't get better either, but that's more of a "shareholder value" problem than it is a specific technology problem.

And yes, I only talked about automation, but the same high-level issues apply to LLMs, but with different downsides: you need to check the LLM output which becomes a bigger topic, and then potentially your own skills stagnate as you rely on LLMs more and more.

[BeetleB]

My first engineering job was non-SW, and had a lot of manual work. I automated a lot of it.

Yes, it led to more work. What would take half a day could now be done in an hour. So we now had to produce 4x more.

I spent 4 years there automating left and right. Everyone silently hated me. One of the problems with my automation was that it allowed for more and more Q/A. And the more you check for quality issues, the more issues you'll find. Suddenly we needed to achieve 4x more, and that meant finding 4x more problems. The thing about automation is that it doesn't speed up debugging time. This leads to more stress.

One senior guy took me aside and said management would not reward me for my efforts, but will get the benefit of all my work.

He was right.

Eventually, I left because I automate things to make my life easier. If it's not making my life easier (or getting me more money), why should I do it?

Since then, whenever I get a new job, I test the waters. If the outcome is like that first job, I stop working on process improvements, and look for another job.

[beachy]

I read a great article a while ago (can't remember where) when they tasked some embedded guys with building a somewhat complex front end app.

When it was done, there were no bugs. Not a single issue. They asked the embedded guys how they had accomplished it. They said "we didn't know bugs were allowed".

Many people have never authored or even been involved with a high quality piece of software, so they just don't know what it looks like, or why you'd want it.

You'd think that someone in the exec team would have some personal pride and ownership in the code and would want to flush out bugs and improve quality. But nah.

[daheza]

This nails so much of my frustration with software development at the moment.

The requests to my team are:

build what product says

close out 90% of the defects you find by priority order

deliver in the priority of feature > security > accessibility

once delivered move on to something else we only have time to work for 3 months on an initiative before we move on

These requirements don't end up with a well working product. They end up with gaps in product, defects that are obvious, non-accessible site. Things take time to polish and be made right, but that's not what is requested. Wanting to iterate and measure isn't important because its not more features.

[EGreg]

One would think that machines and automation would be the perfect thing to catch bugs.

We already do that on many levels -- compilers, linters, pre-commit hooks etc. Well, AI can just red-team and create new tests. The great thing about red-teaming vs blue teaming is that false positive and hallucinations don't hurt the final product. So you can let it go wild.

[swat535]

I mean they don't care about bugs because it doesn't affect sales or bottom line..

Everyone is used to terribly software with awful security holes and performance so why rock the boat?

It's not like average normies will complain.. after all, they are probably used to swimming in an ocean of ads, telemetry and junk. Investors will cut the execs a check regardless, the only thing that matters is piling on more features and growing.

There are _zero_ consequences for writing bad software.

[unloader6118]

Honestly, firmware is usually where we find the worse kind of bugs.

[Nextgrid]

The trick is to automate for your own benefit and keep quiet. Automate your 8hr/day job down to as low as possible, and use the free time for entertainment or another job (where you ideally do the same).

The reward for good work is more work. If they company wanted to pay you more, they would've already done so. If the company wanted automation, they would put that as a job description and pay accordingly (or more likely outsource it and get a shitty result for 10x the price - despite never willing to pay you anywhere close to that even if you were to give them the fully working solution).

[nonethewiser]

This is just the reality of scaling. Largely but not necessarily automation. Think of customer service now compared to early 2000s. Thats not really a story of automation. Instead, it's a story of 1) outsourcing 2) a bit of legitimate self service options (automation) and 3) abandonment - they simply stopped supporting at a good level. Quality is much worse but throughput is much higher - a necessary evil to scale.

AI actually has some ability to improve things. At least when I think about manufacturing and farming. When you produced at such a massive scale you could never individually inspect every potato, widget, or target every weed etc. You could produce WAAAY more but more bad products went out the door. But now you can inspect every individual thing. May not extend to every industry though.

[donatj]

I have a friend who automated his entire days work down to the click of a single button. He did not tell management because they were pretty scummy. He got written up for not "looking busy" despite his output being higher than his coworkers.

Business is stupid. They value busy-ness over productivity.

[BeetleB]

> He got written up for not "looking busy" despite his output being higher than his coworkers.

Also my experience with that first job. I would get the work done quicker than others, and leave around 5pm (most stayed beyond 6pm).

The message was clear: "There's always work to do. If you're getting work done early, you need to do more!"

I got worse ratings than people who achieved less. It also explains why coworkers refused to learn how to automate things.

Again: I automate to make my life easier. If it isn't working, I shouldn't do it.

[overfeed]

> Business is stupid. They value busy-ness over productivity.

Not stupid, just entitled to all of your innovation and productivity while you're on the clock (if waged) and off the clock (if you're salaried). If you've shown yourself to be an outlier - that's great for the business - and congratulations, you've aet yourself a new baseline. Isn't class economics just delightful[1]?

The only employees who have a more direct linkage between productivity and income are sales folk, and it's boom or bust there. If you're an engineer that somehow doubles your employers profits, don't dream they'll double your salary, a once-off bonus is the best you can hope for, at the next evaluation cycle.

1. From each, according to his ability. To each, according to "market" rates, and his negotiation skills.

[popoflojo]

But that is stupid. If they incentive innovation and productivity gains they could do more business. They don't have to give you the full value of your improvements to incentivise you.

[anonymars]

I guess it is right there in the name, isn't it?

[Aurornis]

Automation is a broad topic. At home I save time because my dishwasher automates washing my dishes. The clothes washer and dryer are a lot easier than doing it by hand. The fruit and vegetable at the grocery store are a lot cheaper than they would be without automation.

I think individuals who get comfortable in their jobs don’t like automation arriving at their station because it upends the order of things just as they were feeling comfortable and stable. Being adaptable now is more important than ever.

> Products don't get better either, but that's more of a "shareholder value" problem than it is a specific technology problem.

This is broadly false. Your laptop is unquestionably better because it was constructed with the help of automated CNC machines and PCB assembly as opposed to workers manually populating PCBs.

Some companies can try to use automation to stay in place with lower headcount, but they’ll be left behind by competition that uses automation to move forward. Once that leap happens it becomes accepted as the new normal, so it never feels like automation is making changes.

[everforward]

> Your laptop is unquestionably better because it was constructed with the help of automated CNC machines and PCB assembly as opposed to workers manually populating PCBs.

This is a fundamentally flawed analogy, because the problems are inverted.

CNC and automated PCB assembly work well because creating a process to accurately create the items is hard, but validation that the work is correct is easy. Due to the mechanics of CNC, we can't manufacture something more precise than we can measure.

LLMs are inverted; it's incredibly easy to get them to output something, and hard to validate that the output is correct.

The analogy falls apart if you apply that same constraint to CNC and PCB machines. If they each had a 10% chance of creating a faulty product in a way that can only be detected by the purchaser of the final product, we would probably go back to hand-assembling them.

> Some companies can try to use automation to stay in place with lower headcount, but they’ll be left behind by competition that uses automation to move forward.

I suspect there will be a spectrum, as there historically has been. Some companies will use AI heavily and get crazy velocity, but have poor stability as usage uncovers bugs in a poorly understood codebase because AI wrote most of it. Others will use AI less heavily and ship fewer features, but have fewer severe bugs and be more able to fix them because of deep familiarity with the codebase.

I suspect stability wins for many use cases, but there are definitely spaces where being down for a full day every month isn't the end of the world.

[jononor]

I would not call validation of PCBs easy conceptually. We had to develop a lot of automation in validation to enable today production. Optical verification has been standard for many years, and X-ray is now getting commonplace also. Flying probes commonly used. Functional automated testing is standard for any non-hobby product. But you are right in that the automation overall was/is bottlenecks on the ability to do QA - which not only prevents defects from being sent out, but is also critical to systematic improvements in the production process (both tuning and new iteration). And I believe you are right to call out LLM based systems to be weak in this area, and it is a limiting factor. I believe that automated QA will be more and more critical to positive LLM impact.

[throwaway31131]

Validation that a PCB was manufactured correctly is... easy. Disagree, but how about VLSI. It's hugely automated. Moore's Law is exponential but team sizes aren't. That productivity gap is made up for with huge amounts of automation. And nothing is easy about manufacturing validation of an ASIC.

I do think one primary difference between physical objects and software is we bother to have precise specifications that one can validate against, and I think that's what you're trying to get at. If all software had that then software could have an "easy" validation story too, I suppose.

I have mixed feelings about precise specifications in software. On the one hand the hardware engineer in me thinks everything should have an exact specification. On the other hand, that's throws away the "soft" advantage which is important for some types of software. So there is a spectrum.

[godelski]

FWIW I don't think there's anything factually wrong with what you said, but I think misses the parent's point. They would be incredibly naïve to say that hardware is easy. But I think they were using "easy" as a relative word, not absolute. As is natural in these conversations, but also easily leads to misunderstanding.

  > I do think one primary difference between physical objects and software is we bother to have precise specifications that one can validate against

Having been on the hardware side and now on software (specifically ML) this is one of the biggest differences I've noticed. It's a lot harder to validate programs. But I think the part that concerns me more is the blasé or even defensive attitude. In physical engineering it often felt "it's the best we can do for now" with people often talking about ideas and trying to make it work. It seemed of concern to management too. But in software it feels a lot more like "it gives the right output" and "it passes the test cases" (hit test cases aren't always robust and don't have the same guarantees as in physical design) and call it done. The whole notion of Test Driven Development even seems absurd. Tests are a critical part of the process, but to drive the process is absurd. It just seems people are more concerned with speed than velocity. A lack of depth, and I even frequently see denial of depth. In physical it seems like we're always trying to go deeper. In software it seems like we're always trying to go wider.

This isn't to say that's the case everywhere, but it is frequent enough. There's plenty of bad physical engineering teams and plenty of great software teams. But there's definitely differences in approaches and importantly differences in thresholds. The culture too. I've never had a physical engineer ask me "what's the value?", clarifying that they mean monetary value. I've had managers do that, but not fellow engineers. The divide between the engineering teams and business teams was clearer. Which I think is a good thing. Engineers sacrifice profit for product. Business sacrifices product for profit. The adversarial nature keeps balance

[godelski]

Be careful of Lemon Markets[0]. The problem with them is that they create a stable low quality state. They tend to happen when product quality is not distinguishable at time of purchase.

Which I think we already see a fair amount of this in tech. Even as very tech literate people it can be hard to tell. But companies are definitely pushing to move fast and are willing to trade quality for that. If you're trying to find the minimum quality that a consumer is still willing to pay for, you're likely in a lemon market.

I mean look at Microsoft lately. They can't even get windows 11 right. There's clear quality control issues that are ruining the brand. Enough that us techies are joking that Microsoft is going to bring about the year of Linux, not because Linux has gotten better (also true) but because Microsoft keeps shooting itself in the foot. Or look at Apple with the new AirPods, they sound like shit. Same with Apple intelligence and liquid glass. A big problem (which helps lemon markets come into existence and be stable) is that competition is weak, with a very high barrier to entry. The market is centralized not only because the momentum and size of existing players (still major factor) but because it takes a lot of capital to even attempt to displace them. That's probably more money and more time than the vast majority of investors are willing to risk and the only ones with enough individual wealth are already tied to the existing space.

I think you also have it exactly right about LLMs and AI. A good tool makes failures clear and easy to identify. You design failure modes, even in code! But these machines are designed for human preference. Our methods that optimize for truth, accuracy, and human sounding language simultaneously optimize for deception. You can't penalize the network for wrong outputs if you don't recognize they are wrong.

A final note: you say velocity, I think that's inaccurate. Velocity has direction. It's more accurate to say speed.

[0] https://en.wikipedia.org/wiki/The_Market_for_Lemons

[everdrive]

>Being adaptable now is more important than ever.

I do actually plan on getting old, and as much as I would love to retire before I'm no longer adaptable, I'm not so sure my finances or my brain will comply.

>At home I save time because my dishwasher automates washing my dishes.

I don't think this fits my analogy, because you personally can go watch TV or read a book or exercise given the time that is saved by the dishwasher. At work, you must be at work doing something else, and the "something else" is seldom a real improvement. If I could automate my job and then go on a hike I'd be a lot more excited about it.

[Aurornis]

> At work, you must be at work doing something else, and the "something else" is seldom a real improvement. If I could automate my job and then go on a hike I'd be a lot more excited about it.

When you find an employer that is happy to pay people to not work, let me know because I also want to work there.

[Nextgrid]

The key is to never let them know you're not working. Deliver the output they want. Whether you personally created that output or a machine is irrelevant.

Labor, just like any market relies on information asymmetry. Your company is in business because it manages to sell something at a higher price than the cost it incurs producing it. Your company will absolutely not give away its "secret sauce" to their customers so they can go off and do it themselves and stop paying.

You should act the same; if you have "secret sauce" that allows you to deliver the expected output quicker, enjoy the free time or put it to use elsewhere.

[dingnuts]

it's called being the employer. if you own the capital and you automate the labor then it's your call what to do with the extra time

[BeetleB]

> When you find an employer that is happy to pay people to not work, let me know because I also want to work there.

This was most employers during COVID :-)

I worked fewer hours, and still got more done than most of my team. Since I didn't come to office, no one knew. As long as I responded to emails/messages in a timely fashion, no one cared.

[bravetraveler]

I fail to see the problem! "Time to lean, time to clean" is fine for someone billing/paid by the hour.

As someone on a salary, when the work is finished... I am too. What's overtime? I believe some paperwork had the word 'exempt' on it. My unvested shares are an incentive to save the place from immolation over the next N years. Where's this 'must be at work doing something else' in the contract, again?

"Where's the loyalty?" I hear someone ask. It passed with a family member and employers that had no compassion.

All this to say, I fully support your testing of the water. It's a strategy I've picked up/adapted, too. The poster above should enjoy the time saved by automation/hike. I shitpost.

[godelski]

The problem is as soon as everyone returned to office they did care. Even while remote many employers acted like they were being cheated because employees would work less or distribute their work throughout the day.

We have a tendency to scream crisis while stock prices and market caps rapidly rise. Every little downturn is evidence for the cry, but that doesn't change the trend. They keep saying that the share holders are the real customers and they seem to be doing perfectly fine regardless of if it's a hiring spree or firing. Regardless of if it's even a global pandemic.

There's 4 companies worth more than $3T, one more than $4T. 11 are worth more than $1T. It's only been 7 years since we broke that $1T barrier. Most of the growth has happened recently too. Even Apple has had bigger swings since the pandemic.

Idk, I don't think these companies are in trouble anywhere near what they claim. More concerning is this rapid growth in value without corresponding game changing products. Sure, we got AI but it hasn't changed the game like the iPhone did. I'd give up AI a lot sooner than I'd give up my smartphone, even if all it did was make calls, play music, and have a web browser. A pocket computer is very handy

[Aurornis]

I saw a very different perspective. Some of the people who got to WFH for the first time thought they were getting away with working less because they could bang out Slack replies on their phone when the notification bell came in, but it was really rather obvious that many weren’t working as much.

I’m still salty about it because the people who played this game poisoned the WFH situation for the rest of us who didn’t use COVID as an excuse to work less and try to pretend we were working more.

[bravetraveler]

> At work, you must be at work doing something else

Speak for yourself, salary means I'm done when the work is. I encourage you to enjoy the hike, book, whatever. That said, I truly hate the induced demand LLMs offer.

[sojournerc]

That works short-term. Long-term, expectations of productivity catch up, and you either deliver more or get laid off. It's a treadmill not a mountain.

[bravetraveler]

Eh, it's worked well for a couple decades. Pointed effort beats toil, every time. Layoffs are like the tide, do you like the beach?

[sojournerc]

I thought the same... Then I got laid off. It can happen, not certain it will happen to you, and delivering quality certainly matters more than loc or stupid metrics. Glad you're in a good situation

[pmg101]

Let's say you could automate your job and go on a hike. Great! You can have a fun hike. But you wouldn't get paid for that.

I think it's broadly reasonable that you would only be paid for doing something someone else needs doing.

[thewebguyd]

> But you wouldn't get paid for that.f

Maybe you wouldn't, but you definitely should. Knowledge workers aren't paid for their labor (in the form of me trading my time and effort for wages), knowledge workers are paid for impact. I'm trading my ability to reason, decide, and create value for the company.

I'm valuable not because I sit at a desk and type for 8 hours. I'm valuable because the outputs of my thinking help move the company forward. My employer isn't buying 8 hours of my time , they're buying the outputs that come from expertise and judgement.

So if I automate something, the company still receives the same value the pay me for whether I perform the task manually or build something that automates it. I work in ops, so if I use ansible and a script to automate patching 100 servers instead of doing it by hand, my employers gets the same result: patched systems. The automation didn't diminish my contribution, it proved it. I get paid the same either way.

In essence, my salary is a retainer. It's payment to keep my expertise availalbe, and working for my employers instead of someone else. It's not payment for activity or time.

[lotsofpulp]

These are contradictory claims:

>In essence, my salary is a retainer. It's payment to keep my expertise availalbe, and working for my employers instead of someone else.

>It's not payment for activity or time.

If the latter statement is true, then you must not have any mandatory hours to be present.

If you do have mandatory hours to be present, then the latter statement is not true.

[ponector]

>> My employer isn't buying 8 hours of my time , they're buying the outputs that come from expertise and judgement

I'm pretty sure your typical managers don't think so.

[anigbrowl]

There are a few managers who think the same way, but not that many.

[throwaway0123_5]

I think the problem is if/when AGI enables "someone else" to not need human employees for ~anything. The people that own physical capital (land, farms, mines, etc.) would have robots and GPT-N to extract value from it. The people who survive based on their labor are SOL. I think it is reasonable that many people won't be excited about that kind of automation.

[nahuel0x]

The problem is capitalism, not automation.

[throwaway0123_5]

I don't disagree.

Social/economic stratification (to a certain degree) makes sense as long as there is a reasonable amount of social mobility. AGI paired with advanced robotics seems as though it would all but eliminate social mobility. What would your options be? Politics, celebrity, or a small number of jobs where the human element is essential? I think the economic system needs to dramatically change if/when we reach that point (and ideally before, so people don't suffer in the transition).

[everdrive]

Of course, but then why would I be excited about automation? I can imagine that the executives and shareholders could be excited for automation, but I'm not sure that it benefits me whatsoever.

[lotsofpulp]

Automation reduces cost of goods sold, so in a market with multiple sellers, it leads to lower prices.

Also, almost everyone is a shareholder, directly or indirectly by being a taxpayer and shouldering the cost of pensions, which are invested in businesses.

[Coffeewine]

The only advantage is that if the company is more efficient they'll be less likely to fire you because the business is failing. They'll just be firing you to eliminate a cost.

[lotsofpulp]

When a buyer shops at a lower priced store, they are also eliminating a cost. No one seems to bemoan that, but for some reason a buyer of labor qualified as “employee” eliminating costs is different than a buyer of say, a new roof shopping around or going to Costco to spend less than the full service grocery business.

[bcrosby95]

The executives and shareholders will only be excited about the first order effects of widespread automation like this.

They will be less excited about the second order - a steady loss of revenue as whole professions are automated and people can't find a well paying job.

The third order will be even worse when no one has a job or money to buy anything.

People always point to the industrial revolution. But that created millions of jobs before it obsoleted millions of jobs - you needed workers to create tractors. This wave seems to be shaping up much more like what happened to the rust belt in the late 20th century, regions which still haven't recovered. However this time it'll hit pretty much everyone, everywhere.

Good luck with that capitalism.

[pmg101]

I concur capitalism has it's problems but if this means we move back to feudalism I think we can safely say that will be worse.

[ponector]

>> you would only be paid for doing something someone else needs doing

Right, like drinking coffee at the kitchen in the office.

[candiddevmike]

I think "bottom up" or worker led automation works far, far better than top down. Leadership always comes up with "efficiency" ideas for automation without ever spending a day in the life of the people who will use the automation. And they almost always fail to realize any gains but disrupt everyone's workflow.

[lovich]

> I think individuals who get comfortable in their jobs don’t like automation arriving at their station because it upends the order of things just as they were feeling comfortable and stable. Being adaptable now is more important than ever.

Look at all the other threads with people’s experiences. They aren’t unhappy with automation because they were comfortable. They are unhappy with automation because the reward for being more productive is higher expectations and no compensation.

People think the Luddite movement was smashing looms because they inherently hated technology. They smashed the looms because the factories were producing more and the result of that productivity was the workers becoming destitute.

If the machines and progress only bring about a worse life for individuals, those individuals are going to be against the machines

[edflsafoiewq]

You recoup the saving of home automation immediately as additional leisure time. But for most people, work automation neither reduces your working time nor increases your wage.

[jononor]

Are you sure you spend less time per week on clothes washing than people did before the clothes washer? I suspect that before, people would wash their clothes considerably more seldom - which might even out the efficiency gains of the machine.

[jadelcastillo]

It's an interesting analogy. But one difference between dishwashers and LLMs is that you don't need to check the dishes afterward (if you maintain and use it properly).

[almosthere]

Yeah but to continue the analogy, the washer was JUST invented and your clothes will come out ruined for a while.

[fragmede]

Oh man, remember how much bigger a deal it was that you had to separate your clothes into the exact right categories and run the machine with different kinds of loads? Modern detergent, it's basically all machine wash cold, with far fewer exceptions compared to 30 years ago.

[AlexandrB]

I don't think it's just the detergent. Modern clothes are made of shitty, synthetic fabrics. This is also why most people don't have to iron anything anymore. The tradeoff is microplastics[1], comfort/breathability, and durability.

Classic example is jeans. Modern jeans are ridiculously stretchy compared to "real" cotton denim because they contain tons synthetic fibers. However I run through jeans at an alarming pace - even compared to when I was a kid. They wear quickly, tear easily, and generally don't last.

[1] https://www.pbs.org/newshour/science/laundry-is-a-top-source...

[denimnerd42]

im not sure how youre getting around the physics of white lint on dark clothing. and i separate plastic clothing from cotton or wool clothing because cotton usually just gets machine dried but i dont want to put my wool or plastic clothing through that process because its not necessary.

[anigbrowl]

Why are you putting your clothes in the dishwasher

[almosthere]

Oh crap, maybe that's why they are coming out ruined.

[jadelcastillo]

True, but reaching intelligence is more complicated than cleaning some spoons.

[fragmede]

The two other things that come immediately to mind are clothes; if a shirt cost $4,000 per, our closets would look way different, and cars. No matter your personal opinion on cars vs public transportation, if even if the cheapest vehicle cost $500,000, society would look way different. The real thing it exposes though, is which side of the capital vs labor you work on. If the widget factory suddenly is able to make 10x the widgets in the same amount of time thanks to a new automated widget machine, if you're capital, you now have 10x the widgets to sell. Awesome! However, if you're labor, you still have a 40/hr a week job, regardless of how many widgets you make in a week. And the boss is counting how many widgets you make on the new machine they bought. At the edges of this in the tech industry we have website building. The market haven't yet totally adjusted to the lower costs of labor. What used to take 10 hours to build and you'd charge a client $3,000 for, now takes 2 hours but since the client was previously paying $3,000 for that service, you're not going to charge them less, you're going to take on additional clients. Or spend more time at the beach. In this scenario, the programmer is capital, not labor, and gets to reap the rewards of automation. Until the market catches up, anyway. Given that the industrial machine in the website builder's factory is a laptop and a cloud hosting bill, it's unclear if the Marxist division between capital and labor, burgousie and proletariat is still the right place to draw the lines, but the trade off is still there. If you're selling your time in exchange for money, automation means a faster conveyor belt that you need to adapt to, but you're still working 40/h a week. If you're selling widgets, automation means more widgets to sell.

[rightbyte]

Petite bourgeoisie maybe?

[harvey9]

The customer who was paying 3000 for a website may now be going to somewhere like Wix.

[fragmede]

That client was going to use Wix before AI. B2B sales is a whole thing, and businesses pay to have a human to talk to, not a website.

[microtonal]

I think it all hinges on recognizing what opportunities automation helps.

For instance, I had to rename a collection of files almost following a pattern. I know that there are apps that do this and normally I’d reach for the Perl-based rename script. But I do it so irregularly that I have to install it every time, figure out how I can do a dry run first, etc. Meanwhile, with the Raycast AI integration that also supports Finder, I did it in the 10-15 seconds that it took to type the prompt.

There are a lot of tasks that you do not do often enough to commit them fully to memory, but every time you do them it takes a lot of time. LLM-based automation really speeds up these tasks. Similar for refactors that an IDE or language server cannot do, some kinds of scripts etc.

On the other hand LLMs constantly mess up some algorithms and data structures, so I simply do not let LLMs touch certain code.

It’s all about getting a feeling for the right opportunities. As with any tool.

[brendoelfrendo]

> For instance, I had to rename a collection of files almost following a pattern. I know that there are apps that do this and normally I’d reach for the Perl-based rename script. But I do it so irregularly that I have to install it every time, figure out how I can do a dry run first, etc. Meanwhile, with the Raycast AI integration that also supports Finder, I did it in the 10-15 seconds that it took to type the prompt.

> On the other hand LLMs constantly mess up some algorithms and data structures, so I simply do not let LLMs touch certain code.

See, these two things seem at odds to me. I suppose it is, to a degree, knowledge that you can learn over time: that an LLM is suitable for renaming files but not for certain other tasks. But for me, I'd be really cautious about letting an AI rename a collection of files, to the point that the same restrictions apply as would apply to a script: I'd need to create the prompt, verify the output via a dry run or test run, modify as necessary, and ultimately let the AI loose and hope for the best.

Meanwhile, I probably have a script kicking around somewhere that will rename a batch of files, and I can modify it pretty quickly to match a new pattern, test it out, and be confident that it will do exactly what I expect it to do.

Is one of these paths faster than the other? I'm not sure; it's probably a wash. The AI would definitely be faster if I was confident I could trust it. But I'm not sure how I can cross that threshold in my mind and be confident that I can trust it.

[saxenaabhi]

> See, these two things seem at odds to me. I suppose it is, to a degree, knowledge that you can learn over time: that an LLM is suitable for renaming files but not for certain other tasks. But for me, I'd be really cautious about letting an AI rename a collection of files, to the point that the same restrictions apply as would apply to a script: I'd need to create the prompt, verify the output via a dry run or test run, modify as necessary, and ultimately let the AI loose and hope for the best.

Why? I never understand this level of caution since don't we all use VC? Just feed it the prompt and if it messes up undo the changes.

[acuozzo]

> don't we all use VC?

This assumes you're working with text files.

What if you're working with ~100MiB (each!) frames from a scan of a 35mm movie?

(Note: This isn't fictional. I've worked with file-sets like this in film restoration many times.)

[brendoelfrendo]

As another commenter suggested, this only works for some workflows. I'd also argue it kind of undermines the idea that an LLM can do this work better than a script.

[gmadsen]

as part of the prompt, have a test suite with test files. Its still fully automated by the LLM but adds confidence

[recursive]

If it's under the umbrella of LLM automation, then I'd also need to verify that the test suite behavior actually matches the "production" behavior.

[gmadsen]

sure, but that is less work. you can also have separate LLM QA prompts that assess test suite behavior to production behavior.

ultimately you are right, the buck needs to stop somewhere, but at least in my experience, the more you add quality/test checks as LLM workflows, the higher the rate of success.

[surajrmal]

Do you like washing laundry at the river or carrying water from the well back to your house? You cannot talk in generalities about this topic as it is too broad.

There are definitely many things which when automated loses out on some edge cases. But most folks don't need artisanal soap.

[bdangubic]

> and then potentially your own skills stagnate as you rely on LLMs more and more.

I hear this so often these days and I quite do not understand this part. If I trust LLM do to "X" that means i have made a determination that LLM is top-notch with "X" (if I did not make this determination then letting LLMs do X would be lunacy) and henceforth I do not give a flying hoot to know "X" and if my "X" skills deteriorate it is same thing as when we got equipment to tend to our corn fields and my corn picking skills deteriorated. of course I am being facetious here but you get the point.

[HardCodedBias]

Automation increases productivity.

Without automation we would all be living in poverty.

[kogasa240p]

IMO it's an economic problem (GDP must always rise because reasons) and the hedonic treadmill at play. I would even argue against the complexity point and rather point to overengineering being the root cause; an example would be using a robot arm to automate a cup of coffee instead of using existing vending machines for that purpose.

> potentially your own skills stagnate as you rely on LLMs more and more.

There were some papers from microsoft that highlighted this point https://www.microsoft.com/en-us/research/wp-content/uploads/...

[jstummbillig]

This is just empirically not true. Increase in productivity has lead to broad increase in prosperity (with one relatively recent but currently very important caveat, the housing market).

Now, if what you actually want is to be relatively more prosperous and have more status that's a game you can keep playing forever. But you really don't have to, to simply be better off than all people in the past with far less work.

[subsection1h]

> Increase in productivity has lead to broad increase in prosperity

All of my grandparents retired in their 50s with fat pensions and then lived into their late 80s without having ever stepped foot on a college campus.

[jstummbillig]

You can do that today. But there was a no episode in history where that would have bene the norm or more likely than it is today. Anecdotes are just that.

[lovich]

Where?

The only place I can think of giving pensions at that age anymore is the military. And you aren’t getting a fat pension without being an officer which requires a degree

[philipwhiuk]

A bigger caveat is that measuring improvement by 'prosperity' is both vague (are you using GDP, GDP/capita or GDP/capita of the lowest 10%) and arbitrary (perhaps a better measure is the life expectancy of the poorest 10%).

[jstummbillig]

That does not seem like a caveat at all, given that the improvement is completely obvious for all of these.

[lovich]

> Now, if what you actually want is to be relatively more prosperous and have more status that's a game you can keep playing forever. But you really don't have to, to simply be better off than all people in the past with far less work.

Everyone I grew up with or met via work that is my age or younger has 1-3 more degrees than their parents and grandparents and are significantly worse off when it comes to standard life milestones like buying a home or ever having children.

We are not becoming relatively more prosperous as a people. We have more bread and circuses and less roofs over our heads on average

[j45]

Premature automation is what causes problems.

Too many people are trying to jump to the end when they don't even have their day to day managed or efficient today can tend to carry forward efficiency in a number of business workflows.

Checking the LLM output is required when it's not consistent, in many cases maintaining the benefit requires the human to know more on the subject than the LLM.

[Xevion]

Lowkey it kind of sounds like capitalism is the problem here, not automation.

[pdntspa]

This needs to be boosted more to this community; WE are creating the tools of our own oppression.

The folks at the top know how susceptible we are to being nerd-sniped and how readily we will build these things for them.

[sothatsit]

I want to create automation and greater efficiencies because I believe it is good for the world to have better goods and services, cheaper.

The bigger problem I see is not automation, it is the exploitation of addictive behaviours to “capture attention”.

[cool_man_bob]

I don’t give a fuck how cheap you make bread and circuses. The only goods and services I give a fuck about anymore are the ones that won’t be made cheaper with automation.

[sothatsit]

Speak for yourself. I want cheaper building costs so we can build more housing, cheaper and safer vehicles, higher quality food so we can all be healthier, better medical technology and medicines so we can solve more diseases, and new washing-machine-like technologies so I can spend more time with friends/family. That’s not to mention that greater leverage on my labour would give me even more flexibility to choose the work I want to do, and how much I want to work.

Bread is already so cheap as to not notice the price most of the time. But other goods and services are absolutely not that cheap. And there’s certainly higher quality that could be achieved, especially in areas like medicine. It is a lack of imagination to not see all the ways in which cheaper goods and services could improve our lives.

[throwaway0123_5]

Making all of those things cheaper is great, as long the automation isn't also making everyone poorer at an equal or faster rate. It doesn't really help if house prices and food prices are cut in half if most people lose their employment because of automation.

I think the concern is that true human+ AGI and advanced robotics would obsolete so many roles that it doesn't matter if things can be made more efficiently, because nobody will have any money at all. If/when AI can do my job better than me, it isn't giving me leverage, it is removing all leverage I have as someone who puts food on the table through labor.

In the interim period before that happens then sure, the automation is great for some people who can best leverage it.

[pdntspa]

It's never going to go down like that as long as companies are required to serve shareholder interests above customers' or employees'.

Instead all these automation tools are and will be used to cut corners and optimize on cost. Quality, peace-of-mind, and increased free time will be the sales pitch used to placate us plebes. But we all know what the executive dipshits will really care about.

[6gvONxR4sf7o]

As always, labor is a marketplace, and the supply side boils down to a) how much the next person else is willing to work (all else equal), and b) external forces (like overtime requirements kicking in at 40 hours).

[mclau153]

You could remove "you still need to work 40 hours a week"

[risyachka]

Its like a factory.

First things were made by hand, slowly - they were expensive and you could make a living making things.

Now those things are made in factories.

And they are 99% automated - like where software is going.

And whats left is to be a mindless factory worker doing repetitive things all day for a living wage.

But hey, you are so productive - now you make 100k items in a day. Must feel nice.