[dakiol]

Why do we assume that Prompt Engineering is going to pay less money? As usual, what one brings to the company is value, and if AI-generated code needs to be prompted first and reviewed later, I don’t see how prompters in the future could earn less than software engineers now.

Prompt engineering is like singing: sure thing everyone can physically sing… now whether it’s pleasant listening to them is another topic.

[cardanome]

The first thing any big technical revolution causes is suffering for a lot of people.

It can bounce back over time and maybe leave us better off than before but the short term will not be pretty. Think industrial revolution where we had to stop companies by law from working children to literal death.

Whether the working man or the capital class profits from the rise of productivity is a questions of political power.

We have seen that productivity rises do not increase work compensation anymore: https://substack.com/home/post/p-165655726

Especially we as software engineers are not prepared for this fight as unions barely exist in our field.

We already saw mass layoffs by the big tech leaders and we will see it in smaller companies as well.

Sure there will always be need for experienced devs in some fields that a security critical or that need to scale but that simple CRUD app that serves 4 consecutive users? Yeah, Greg from marketing will be able to prompt that.

It doesn't need be the case that prompt engineers are paid less money, true. But with us being so disorganized the corporations will take the opportunity to cut cost.

[bgwalter]

> Especially we as software engineers are not prepared for this fight as unions barely exist in our field.

You can fight without unions. Tell the truth about LLMs: They are crutches for power users that do not really work but are used as excuses for firing people.

You can refuse to work with anyone writing vapid pro-LLM blog posts. You can blacklist them in hiring.

This addresses the union part. It is true that software engineers tend to be conflict averse and not very socially aware, so many of them follow the current industry opinion like lemmings.

If you want to know how to fight these fights, look at the permanent government bureaucracies. They prevail in the face of "new" ideas every 4 years.

[jopsen]

> If you want to know how to fight these fights, look at the permanent government bureaucracies. They prevail in the face of "new" ideas every 4 years.

Search youtube for "yes minister" :)

-----

On topic, I think it's a fair point that fighting is borderline useless. Companies that don't embrace new tech will go out of business.

That said, it's entirely unclear what the implications will be. Often new capabilities doesn't mean the industry will shrink. The industry haven't shrunk as a result of 100x increase in compute and storage, or decrease in size and power usage.

Computers just became more useful.

I don't think we should be too excited about AI writing code. We should be more excited about the kinds of program we can write now. There is going to be a new paradigm of computer interaction.

[jplusequalt]

>You can fight without unions.

And you can fly without wings--just very poorly.

Unions are extremely important in the fight of preserving worker rights, compensation, and benefits.

[autoexec]

> You can fight without unions.

You can fight without an army too, but it's a lot less effective. There is strength in numbers. Corporations know this and they leverage that strength against their employees. You all alone vs. them is exactly how they like it.

[StefanBatory]

> You can refuse to work with anyone writing vapid pro-LLM blog posts. You can blacklist them in hiring.

This works only if everyone is on with this. If they're not, you're shooting yourself in the foot while doing job hunting.

[jdright]

> This addresses the union part.

lol, good luck with that.

you thinking that one or two people doing non organized _boycott_ is the same thing as an union tell a lot about you.

[bgwalter]

I didn't mention one or two people, I mentioned a theoretical strategy that is also employed by non-unionized bureaucracies. And I admitted that the strategy is impeded by the fact that many software developers are self-loathing people with no spine.

It is possible that obedient people need highly paid union bosses, i.e., new leaders they can follow.

[jdright]

You're wrong, obedient people are exactly the type that don't need unions, they are obedient and accept anything.

Unions are for people that don't accept anything and know that they are a target taking action alone or in non organized ways.

Unions are the way to multiply the forces and work as a group with common interests, it is for people that are not extremely selfish and egocentric.

[redeeman]

why do you need a union? everyone should just set their personal standards for what they accept/demand, and then let the market sort it out? someone wants to work for $20/hour programming, in a fashion that can satisfy some demand? great, then I will simply NOT be doing that job. Everyone wins even though I do not get that particular job. Someone is willing to work 7 days a week as they prefer to grind to earn more money? good on them. Its not gonna be me, They win the job, I dont.

Nobody wants to inhale toxic fumes in some factory? well then the company had better invest in safety equipment, or work dont get done. We dont need a union for this

[paulcole]

> The first thing any big technical revolution causes is suffering for a lot of people.

Didn’t Greg-from-marketing’s life just get a lot better at the same time?

[trod1234]

> The first thing any big technical revolution causes is suffering for a lot of people.

This all assumes that such revolutions are built on resiliency and don't actually destroy the underpinning requirements of organized society. Its heavily skewed towards survivor bias.

Our greatest strength as a species is our ability to communicate knowledge, experience, and culture, and act as one large overarching organism when threats appear.

Take away communication, and the entire colony dies. No organization can occur, no signaling. There are two ways to take away communication, you prevent it from happening, or you saturate the channel to the Shannon Limit. The latter is enabled by AI.

Its like an ant hill or a bee hive where a chemical has been used to actively and continually destroy the pheromones the ants rely upon for signalling. What happens? The workers can't work, food can't be gathered, the hive dies. The organism is unable to react or adapt. Collapse syndrome.

Our society is not unlike the ant-hill or bee hive. We depend on a fine balance of factors doing productive work and in exchange for that work they get food, or more precisely money which they use to buy food. Economy runs because of the circulation of money from producer to factor to producer. When it sieves into fewer hands and stays there, distortions occur, these self-sustain and then eventually we are at the point where no production can occur because monetary properties are lost under fiat money printing. There is a narrow working range where outside the range on each side everything catastrophically fails. Hyper-inflation/Deflation

AI on the other hand eliminates capital formation of the individual. The time value of labor is driven to zero. There is a great need for competent workers for jobs, but no demand because no match can occur; communication is jammed. (ghost jobs/ghost candidates)

So you have failures on each end, which self-sustain towards socio-economic collapse. No money circulation going in means you can't borrow from the future through money printing. Debasement then becomes time limited and uncontrollable through debt traps, narrow working price range caused by consistent starvation of capital through wage suppression opens the door to food insecurity, which drives violence.

Resource extraction processes have destroyed the self-sustaining flows such that food in a collapse wouldn't even support half our current population, potentially even a quarter globally. 3 out of 4 people would die. (Malthus/Catton)

These things happen incredibly slowly and gradually, but there is a critical point we're about 5 years away from it if things remain unchanged, there is the potential that we have already passed this point too. Objective visibility has never been worse.

This point of no return where the dynamics are beyond any individual person, and after that point everyone involved in that system is dead but they just don't know it yet.

Mutually Assured Destruction would mean the environment becomes uninhabitable if chaos occurs and order is lost in such a breakdown.

We each have significant bias to not consider the unthinkable. A runaway positive feedback system eventually destroys itself, and like a dam that has broken with the waters rushing towards individuals; no individual can hold back those forces.

[RugnirViking]

There are people trying very hard (and succeeding) to CREATE the impression it will pay less money. Pay in general is extremely vibes based. A look at well... anything in the economy shows that. There are constant shortages of low paying jobs, and gluts of high paying jobs

[giantg2]

That's the opposite of what I'm seeing. I see plenty of openings at Walmart, bus drivers, etc. I see very few openings in many higher paying jobs (healthcare might be an exception). Even the dev jobs I'm finding are low paid at small companies, like $70k per year low (and this isn't a low cost area).

[bluGill]

Wait a couple years before you start stating anything as a trend. There have been several downturns over my lifetime (I'm 50) where it was hard to find a tech job, and several periods of good times where tech people were in high demand. Until several years have past you cannot tell the difference between a temporary downturn and the end of an era.

Every time things turn bad a lot of people jump out and yell it is the end of the tech. They have so far been wrong. Only time will tell if they are right this time, though I personally doubt it.

[giantg2]

I'm not saying it won't turn around, but its been about 2-3 years.

[sokoloff]

If it’s been 3 years, then I’m pretty sure it’s not AI-coding driven.

[giantg2]

Copilot has been around that long.

[al_borland]

I suspect it only looks like there are a lot of high paying jobs, because they are so much harder to fill, due to a there being so few qualified candidates... hence the high pay. Supply and demand.

[bluGill]

Supply and demand is also influenced by ability to become qualified. I'm not qualified to flip burgers, but any fast food place could get me qualified in just a few hours. I'm not qualified to be a doctor and it would take me years of training to get that qualification. I've met people who failed to qualify as a burger flipper - you can correctly guess by that statement they are very disabled, I've met many who failed to qualify as a doctor, and all are smart people since people who are not wouldn't even try.

[somenameforme]

People aren't paid by value brought to companies, they're paid by the scarcity of their skill. Your analog is actually perfect for this. There's a reason saying you want to be a professional singer is generally something only a child would say. It's about as reliable a career as winning the lottery, simply because everybody can sing, lots of them quite decently. And so singer, as a career, mostly isn't a thing - it's a hobby with some distant hope of going Oliver Anthony at some point.

Software development has a huge barrier to entry which keeps the labor pool relatively small, which keeps wages relatively high. There's going to be a way larger pool of people capable of 'prompt engineering' which is going to send wages proportionally way down.

[dakiol]

> There's going to be a way larger pool of people capable of 'prompt engineering' which is going to send wages proportionally way down.

My wife knows how to prompt chatgpt, but she wouldn't be able to create an app just by putting together what the llm throws at her. Same could be said about my junior engineer colleague; he knows way more than my wife, sure, but he doesn't know what he doesn't know, and it would take a lot of resources and effort for him to put together a distributed system just by following what an llm throws at him.

So, I see the pool of potential prompters just as the pool of potential software engineers: some are good, some are bad, there will be scarcity of the good ones (as in any other profession), and so wages don't necessarily have to go down.

[somenameforme]

Of course, but again the issue is the number of people. Software development, as it currently is, has huge barriers to entry. Working in code is something that many people simply cannot do, and of those that can - a huge chunk will find it intolerable as an occupation. 'Prompt engineering' will have far smaller barriers to entry which will, even all other things being equal, significantly increase the labor pool.

[sokoloff]

…unless the value created via prompt engineering is high enough to cause companies to rationally demand even more prompt engineers.

The size of the pie is nowhere near fixed, IMO. There are many things which would be valuable to program/automate, but are simply unaffordable to address with traditional software engineering at the current cost per unit of functionality.

If AI can create a significant increase in productivity, I can see a path to AI-powered programming being just as valuable as (and a lot less tedious than) today.

[somenameforme]

Again it's not about value, but solely supply vs demand. If there was somehow only one person who could do janitorial work in a city, that'd be one rich janitor.

For a more realistic example - the software side at many companies essentially is the company. They bring products all the way from inception to launch. Yet they tend to get paid less, often much less, than the legal side. The reason is simply that the labor pool for lawyers is much smaller than for software engineers.

If there's not significant barriers to entry for prompt engineering, wages will naturally be low.

[sokoloff]

Demand incorporates/expresses value (fairly obviously).

[somenameforme]

Only if one uses circular logic when defining it, which is I find the MBA style definition of value to be mostly meaningless and also misleading, because people interpret it as they want. By contrast directly speaking of supply/demand encompasses everything quite well and eliminates any circular logic, let alone the need for such.

[sokoloff]

How would the demand curve avoid expressing anything the value the buyer receives (or at least perceives) from their transaction?

[bgwalter]

Singing properly requires decades of training. Prompt engineering is like a 5 year old asking his parents for an ice cream. Some strategies are more successful than others.

[jplusequalt]

>Why do we assume that Prompt Engineering is going to pay less money

It objectively takes less expertise and background knowledge to produce semi-working code. That lowers the barrier to entry, allowing more people to enter the market, which drives down salaries for everyone.

[OkayPhysicist]

The best paying jobs are ones that look inapproachable and are inapproachable. (even if that's for two entirely different reasons). Most people look at surgeons, for example, and go "I could never do that", and their jobs are, in fact, difficult and require lots of training, and the combination of those two factors primes people to be willing to give them a bunch of money. Jobs that look hard but are (compared to appearances) easy also tend to pay pretty well.

But jobs that look easy or approachable are in a much tighter spot. Regardless of how difficult they actually are, people are far less willing to give them large amounts of money. Pretty much all the more artistic jobs fall into this camp. Just because any idiot can open up Photoshop and start scribbling, it doesn't follow that competent graphic design is easy.

Right now, software development is incidentally in the "looks hard is hard" category, because the reason it "looks hard" is entirely divorced from the reason it is hard. Most of the non-tech population is under the obviously (to us) incorrect impression that the hard part of programming is understanding code. We know that that's silly, and that any competent programmer can pick up a new programming language in a trivial amount of time, but you still see lots of job postings looking for "Java Developers" or "Python Developers" as opposed to actual domain specific stuff because non-technical folk look at a thing they know is complicated (software development) see the first thing that they don't understand (source code) and assume that all the complexity in the space is tied up in that one thing. This is the same instinct that drives people to buy visual programming systems and argue that jargon needs to be stripped out of research papers.

The shift over to plain-language prompt engineering won't solve the underlying difficulty of software development (requirement discovery, abstract problem solving), but it does pose the threat of making it look easy. Which will make people less prone to giving us massive stacks of money to write the magic runes that make the golems obey their commands.

[blindriver]

Because the 10% difference between the best prompt engineer and a mediocre prompt engineer won't usually make a noticeable difference in output or productivity. There's no reason to specialize in it or pay extra because the gains are ephemeral.

[sokoloff]

That’s a wild take to me.

I think that the spread of capability and effectiveness between the best and mediocre will continue to be several factors and might even increase as compared to today.

I can’t see any way it would be less than 2x.

[blindriver]

Conversely I think that 2x being the least amount of difference between mediocre and best prompts seems unbelievable. I've been using LLMs heavily just like everyone else, and I find that simple direct prompts get me most of what I need. I've never seen it get me less than half of what I expect at all.

[sokoloff]

Think of Google searching circa 2010 (before they completely screwed it up for profit). The best Google search user was way more effective than a mediocre searcher, even on the much simpler task of searching.

I don’t think you’re giving yourself enough credit for understanding enough about system design to be able to effectively prompt and guide the LLM.

John Carmack* is going to be at least twice as good at making a game with LLM coding assistants as a mediocre prompt engineer will be. Guido van Rossum* and Rich Hickey* will be over 2x a mediocre prompt engineer at language creation. Linus Torvalds* will create the first version of git far faster than any mediocre prompt engineer (who will never complete that task). And on and on…

* replace with whomever you think is “best” in the field you’re comparing.

[blindriver]

I guess I disagree with you. I think that results between average and the "best" prompters are getting closer and closer together because that's what LLM designers and customers want.

[sokoloff]

Sure. Everyone has incentive to raise the effectiveness of the bottom of the distribution as a buyer or seller of these tools. That’s where the big payday is, but incentive alignment doesn’t mean that it’s possible to lift mediocre all the way up to sit right next to best.

[giantg2]

If you need fewer prompt engineers than developers to do the same work, and if prompt engineering is easier than developing meaning all developers can do it, then you need up with a massive labor oversupply.

[bapak]

How many programmers did you need 40 years ago to write MS DOS programs? As you become more productive, more is expected of you. Instead of spending 10 days coloring pixels on the screen, now you're expected to push whole UIs in the same amount of time. Whether this is enabled by high-level languages, tools or AI is irrelevant.

[rhines]

I wonder if we need teams generating dozens of UIs or whatever every day though. There may (or may not) be a limit to how much value-adding work is available to do, or at least diminishing returns that no longer justify high salaries.

[bluGill]

YEs we do - your point that there might be too many is valid, but a modern UI when done right is much more accessible to the "common man" than a MSDos and so all the time those teams put in is more than made up for in all the time you saving not having to teach all the people who will use the program and thus we need far more teams than in the MSDos days when we couldn't make a good UI in the first place.

[giantg2]

What am I to be productive on? The bottleneck now is finding a business idea that's viable.

[chii]

> Prompt engineering is like singing

i think you got the analogy wrong. Not everyone can sing professionally, but most people can type text into a text-to-speech synthesis system to produce a workable song.

[lubujackson]

Maybe a better analogy is that now anyone can use autotune now and actually sing, but you still have to want to do it and put in the effort. Very few people do.

[roland35]

Well there's value, but also supply and demand! If prompt engineering is easier that means more people will be able to do it!

[trod1234]

> Why do we assume that Prompt Engineering is going to pay less money.

I supposed because every new job title that has come out in the last 20+ years has followed the same approach of initially the same or slightly more money, followed by significant reductions in workforce activities shortly thereafter, followed by coordinated mass layoffs and no work after that.

When 70% of the economy is taken over by a machine that can work without needing food, where can anyone go to find jobs to feed themselves let alone their children.

The underlying issues have been purposefully ignored by the people who are pushing these changes because these are problems for next quarter, and money printing through non-fraction reserve banking decouples the need to act.

Its all a problem for next quarter, which just gets kicked repeatedly until food security becomes a national security issue.

Politician's already don't listen to what people have to say, what makes you think they'll be able to do anything once organized violence starts happening because food is no longer available, because jobs are no longer available.

The idiots and political violence we see right now is nothing compared to what comes when people can't get food, when their mindset changes from we can work within the system to there is no out only through. When existential survival depends on removing the people responsible by any means, these things happen, and when the environment is ripe for it; they have friends everywhere.

UBI doesn't work because non-market socialism fails. You basically have a raging fire that will eventually reach every single person and burn them all alive, and it was started by evil blind idiots that wanted to replace human agency.