AI and globalisation are shaking up software developers' world
economist.com
economist.com
My friend dubbed this effect "AIrony"
Most of this country checks out their own groceries and AI has started stocking shelves.
Creative industries are becoming commodities due to tooling.
All I can come up with when I search for safe jobs is athletes, politicians, and live music performers.
Anything where the point is specifically human effort rather than the actual result, plus ones that set their own rules.
So yes, athletes and politicians; for the moment also prostitutes (pending better bioprinting, not because it's in either category); but also some art — not all art, just the art where the point is to show off how rich you are; most live music was automated away a century ago, what remains is likely also the category of art where the expense is the point.
However, Transformers movies and equivalent "art as a product" will be made by algorithm or GenAI, if they aren't already.
So also doctors, lawyers, teachers, soldiers and government workers.
It's not that they can't be replaced by AI, they just won't legally be allowed to.
Teachers cannot be replaced by AI if you see them as more than just "knowledge passing". If you think the human component of the teacher-student interaction is fundamental, as I do, teachers won't be replaced, though they may use AI to assist them.
What is the "human component" to teaching? How can't that be solved via AGI? AGI can do human emotions, current AI already can pretty well for most things.
The "human component" to (good) teaching is human beings teaching other human beings to be human. A teacher is not a mechanism for offloading knowledge into a child-shaped container.
You learn that before you enter school, school teaches you how to be less human and more robotic.
That's a very cynical take, and in my opinion, false too.
You can find examples of bad anything everywhere, including teaching. What you describe hasn't been my own experience as a child, nor my silblings', nor my own children.
> You learn [to be human] before you enter school
Hard disagree. You never stop learning how to be human till the day you die, but your most important formative years in this sense occur during childhood and early adulthood.
https://old.reddit.com/r/notebooklm/comments/1fr31h8/noteboo...
* there's disagreement about each letter, and treating them as boolean values rather than continuous.
But I want human teachers behaving truly human to students, not something that can act "disturbingly capable of emotions" or fake it. AGI is nowhere near this (and there's no good definition of what AGI is, anyway).
Again, the human component is an essential part of the teacher-student relationship. Teachers are not simply offloaders of knowledge. Just like I don't want parent-like robots nursing and raising children (raising humans and being raised by them is essential to the human experience), I don't want teacher-like robots "disturbingly capable" of acting as if they have emotions.
Think of it this way: a sociopath [1] is a human that can act as if they had cooperative/empathetic feelings towards other humans, when in reality acting callously in their own self-interest all the time. Would you want a sociopath teaching children? I wouldn't; whether the sociopath can pass knowledge well and can fake as if they had normal human emotions won't sway my opinion.
Real human connection is what makes us human. AIs are not human by definition. Therefore, AIs cannot replace humans in those situations where human connection matters. And we shouldn't want them to either, because otherwise what's the point of existence? Asimov's fictional planet of Solaria was a dystopia, not something to aspire to.
----
[1] or the pop culture understanding of them, anyway. Let's not get caught up in definitions.
On "should we":
I agree completely.
I've previously made a similar example with sociopaths, when people are incredulous as to the risks of an un-aligned AI.
I've also previously averred that the first time we attempt a brain upload, we're likely to know so little about how our minds function that the upload will have every possible mental health disorder simultaneously, including many that we've got no words for because they're incompatible with life.
On capabilities:
> But I want human teachers behaving truly human to students, not something that can act "disturbingly capable of emotions" or fake it.
If you can't tell, does it matter? Students have been learning from increasingly capable simulacra of human intelligence and emotion since no later than the invention of writing, possibly even earlier with art.
LLMs are good enough to be interesting, and good enough to fool a lot of people a lot of the time, even if last year's models weren't passing Turing test scenarios where people were actively looking for them. (But also, in those tests, humans rated other humans as human only 2/3rds of the time: https://arxiv.org/html/2405.08007v1). Although the newer models are generally better, I don't want to claim if they're specifically, at this task, better than the old models — still, that's how close they were on the only formal test I know of, rather than the mere vibes of listening to them synthesising a voice and me being surprised they'll um and err now.
> AGI is nowhere near this (and there's no good definition of what AGI is, anyway).
I made the parenthetical point myself. LLMs, however, are absolutely there.
Think of it this way: neither one of us can tell by the other's words that the other is not an LLM, all we can do is trust because we're on a forum where it would be rude to use one.
> whether the sociopath can pass knowledge well and can fake as if they had normal human emotions won't sway my opinion.
Sure, but you also wouldn't be able to determine if you're looking at a sociopath or someone with normal human emotions.
> Real human connection is what makes us human.
Para-social relationships say otherwise. Fictional characters, let alone the actors portraying them. The electoral popularity of sociopaths in democracies. Shallow demagogues. Art from generations long-dead. These things give the illusion of a real connection — to the extent there's any connection, it's all one-directional.
> AIs are not human by definition. Therefore, AIs cannot replace humans in those situations where human connection matters.
(Parenthetically: would an uploaded human mind, count as human? I don't know).
But it doesn't matter if they're human or not. Dogs and cats often provide a similar connection (both growing up as a companion, and a being to look after as an adult, though the dog owners I know really don't like anyone treating dogs as "fur babies"), as do TV shows: to the extent that the "connection" part matters, it doesn't need to be human; to the extent that the "human" part matters, it doesn't have to be real.
> And we shouldn't want them to either, because otherwise what's the point of existence? Asimov's fictional planet of Solaria was a dystopia, not something to aspire to.
The trouble with fiction is that it's as much of a shallow echo as current LLMs themselves: yes, it can produce an emotional affect, but for every Asimovian Solaria there's a Banks' Culture where the AI have us thrive.
Both are unsound examples upon which to base actual expectations: even if we happen to develop down one or other of those paths, they are written to be interesting not plausible, and the similarity is mere coincidence.
> [1] or the pop culture understanding of them, anyway. Let's not get caught up in definitions.
I tend to agree, though I feel there's an irony here somewhere — shared human connection on the basis of common understanding of a shared meme about the mental processes of a sub-group of humans that's being Othered in the same way as LLMs… but I can't put it into words. And it would be super ironic to ask an LLM to try to do so for me, so I won't.
--
Don't mistake the map for the territory; but also don't forget you can learn the territory from the map.
Any given AI may be a mere simulacra, a P-zombie mimicking our inner world without having one of its own, but that doesn't mean the model isn't good enough to learn from.
I guess I'm indeed arguing more from the "should we?" angle first, though I think human connection does matter and is currently not fakeable (not by LLMs). It's harder to argue my point because people can (and have) formed strong attachments to "AIs" as simple as ELIZA. I don't doubt there are people out there who can form a strong bond to a Tamagotchi "pet". That's really a human trait, to see "someone" where there is none, or very little of it. In a way, AI (or the companies marketing, anyway) has a dangerous potential to hack and exploit this human tendency to attribute feelings or personality to stuff.
To be clear, I'm not disparaging AI tools or saying AI has no place in the classroom as a tool to assist teachers.
I'm arguing that this won't replace teachers -- the original comment I replied to argued the ony reason AIs won't replace teachers is legal, but I disagree -- because teachers are not "teaching tools"; they are much more than that. A teacher is not a machine with coffee and questions as input, and lessons or disciplinarian actions as output.
> But it doesn't matter if they're human or not. Dogs and cats often provide a similar connection (both growing up as a companion, and a being to look after as an adult, though the dog owners I know really don't like anyone treating dogs as "fur babies")
I know you said not to look at works of (scifi) fiction, but have you read Philip K. Dick's "Precious Artifact"? Your mention of dogs and cats brought it to mind; if you've read it you'll know why :)
No, those are all subordinate to the whims of the leaders.
That said, I should have been clearer and said "leaders" rather than simply agreeing with the word "politicans": a military general who seizes power may have a different title, that title doesn't matter, it is the power that is the point.
Right now most of us are in democracies where the government answers to the people, and the people could set any law requiring any given job be done by humans — but that's not automatically guaranteed going forward, many democracies have falled in the past.
Any job that requires general purpose hands and feet to to "real work". Like the thousands of people fixing infrastructure right now everywhere after Helene.
https://eli.thegreenplace.net/2022/asimov-programming-and-th...
Well, until AGI, of course. But then we're probably all toast anyway
Of course, they're going to destroy their companies when they layoff their developers and replace them with under-skilled prompt engineers. But by then, the damage will be done, both among the employers and the employees.
If you have equity in an AI etc company, you'll be flying.
Wealth concentration may increase, barring a revolution.
I think it's made me 3 -> 5x more valuable.
Now, maybe it's done the same for other people as well, but I think my co-efficient was high, and I continue to insist that I've never worked anywhere that had enough resource focused on software development. The demand has always massively exceeded the organisations ability to fund supply.
And, as many have noted lower costs can also boost demand.
Finally, off shore devs are on a longer cycle time for many reasons. If cost onshore are reduced then that cycle time will crush offshore competitive advantage. I don't see any prospect of me getting work from India or Indonesia, but I do see AI as a driver that will reduce work leaving the UK for other places.
This is huge for me, because I can realise ideas and show people how to solve problems whereas bc (before cursor) I was reduced to saying "I'm sure that will work if we do it that way" and hoping that my folks would follow through on the thinking themselves.
Cursor supports me in two ways.
1) It gets things moving - I can put all the boilerplate and infrastructure to build something in place in a few minutes.
2) It removes road blocks. When I get stuck it gives me fixes.
Now I recognise it's all from Stack Exchange, basically, but it's a terrific interface for that knowledge.
What it doesn't do is provide new knowledge or ideas for fixing things. That has to come from upstairs still, but that's never been an issue for me.
You should negotiate a corresponding raise!
I am a terrible programmer but probably could have BSed my way into a jr front end role with enough effort circa 2019.
The reason I didn't is exactly because of right now. I would be unemployed and chasing dev jobs with zero chance of landing.
Instead I was learning Julia today with the help of language models because I can stretch out and explore with not needing a job.
Right now is just the shake out of the pretender software engineers like I would have been.
People like you will be fine. People like me should be doing something else anyway.
Induction from the perspective of the seasonal turkey: https://mashimo.wordpress.com/2013/03/12/bertrand-russells-i...
And either way, talented developers from developing countries are often brought on with visas and end up with salaries close to the industry norm. I’ve been involved with large mergers with overseas dev teams and that’s how it’s always been.
At best, AI is just good documentation. This whole craze is the equivalent of saying doctors won’t be necessary because you can just google how to do x procedure.
Its creating software applications and then signing up to support said software codebase for 100K+ users and agreeing to pager duty and triage management for it for more than a 5-10 years.
Perhaps AI purveyors should focus on automating customer/bizdev support by creating frameworks/workflows for AI pager duty and AI triage management.
We need to move past this type of thinking. Most professional services could be automated and yet they are not for a number of reasons. For developers, the unescapable complexity that keeps creeping into projects if you don't know what you are doing will keep us employed.
If you dont think this will not happen, you have not used cursor.sh with sonnet 3.5. I am working on a startup that is effectively a copy of existing products. My flow goes:
1. Screen shot competitor UX
2. Drop screen shot into cursor.sh, ask it to code it (there will be issues)
3. Screen shot individual components in the above screenshot
4. Have cursor.sh fix each component one by one
5. Go back through and tell it each issue by highlighting the code
6. Iterate a bit
7. 90% of the UI has been copied in one day, with a working backend.
This is with a next.js/supabase app deployed on vercel. For complex B2B saas.
Its over. In a year the models will get so much better
This tech currently just makes each engineer more efficient. There’s an infinite space of problems to solve. I guess we’ll must be digging through that space faster and wider from now on.
Besides, any sufficiently complex field has a lot of business rules that you can't derive from screenshots.
The level of delusion regarding AI makes me want to completely ignore anything that touches it and come back in 10-20 years when people learned something.
Maybe OP meant "all CRUD TODO-like SaaS is toasted".
This gives us the tooling to A.) burn through those faster. B.) solve more complex problems.
If your entire job as an engineer is “take tickets with very clear acceptance criteria and turn into component / basic business logic”, I’d be worried for sure. I was initially worried myself when all this landed two years ago, but using it daily and now building enterprise scale apps with it, I know we’ve got at least another decade or two before it’s able to tackle the full breadth of an engineers role.
It’s over! All websites can now just be slightly tweaked templates. Go home, we no longer need any web dev whatsoever.
It’s not like the bar will raise. No, everyone will expect the exact same level of quality as of 2024 and bloated, AI diarrhea using 200x resources is fine now and will always be fine.
Nobody needs to differentiate and we will all copy each other endlessly and eventually the whole software world will be basically one single product.
Remember how Brick and Mortar was dead because of online shopping? Who knew in 2024 I'd be more likely to head out to an actual store than risk purchasing some sketchy product off Amazon, but I can imagine that if we do eventually actually get to a future where the world is filled with the white noise of SAAS garbage coming out left and right anyone actually interested in mitigating risk for their business is going to learn quickly to err on the side of caution.
If anyone and everyone can emit noise then that becomes the baseline, and people will catch on quick to find those that rise above.
I have seen this said before. Would you show some of these sites that you have knocked out in one day fully working? If not, would you be willing to do the community a service and over the course of a weekend, live-stream or video yourself duplicating this site[0]? Just the publicly available parts.
It would be eye-opening for many I think, and in the process would make you one of the most famous developers in the world (or at least in the US), since at least two very well-known developers/influencers do not think it is currently possible and dunking on them live or on video screen recording would really be a sea-change in our sector. Not to mention, settling once and for all the "LLMs are going to take our jobs" vs. "lol, that are just a fancier autocomplete" argument within the industry.
I am not joking about this or trying to bait you, I would seriously like to know if I am wasting my time with traditional development and really need to re-orient my skills and job prospects. So far, I have seen many people do demos, but none were anything other than relatively simple, error-prone, SPA apps. It is more complex ones, I am interested in. The link below[0] is not that complex for a traditional developer, but it isn't super-simple either. From your comment, it sounds like you have the experience and ability to do it.
Except this analogy breaks down because photography is but a sliver of what is possible with ubiquitous programmability. It's going to be more, different, and somewhat unexpected.
Get ready!
It's like the web which was originally thought to be the means for people to be creating their own web sites rather than consuming TV content, but look how that went.
I guess nowadays there are an elite few professional youtubers making regular content, and teenagers posting TikTok videos, but that's about it.
Everyone is getting accelerated learning and task completion, but what does that mean for the market? I think there are more fundemental things to this change that are needed to be fleshed out. Pre LLM programmers and post LLM programmers will be very different.
Ah yes, software famously loved by users. Microsoft teams and in-car software.
Isn't it obvious that if a job can be done from anywhere that would mean that you are opening yourself to world wide competition when subsidizing your company with your own facilities?
Was it hubris? Do Americans believed that only they can do this coding stuff? Not only that the job market was opened to global competition, it allowed tech and capital transfer. On the grand scheme its great, now everyone does all these things but from workers perspective its very bad idea.
It has been quite some time to see some outstanding American innovation. OpenAI and the generative AI was one but it didn't took much time for China to leap ahead. Maybe its nationalism, maybe its the language barrier but I see a lot of Asian innovations that don't catch much attention in the Anglosphere. American techies pumping Sora on Twitter over some demos when Chinese are having it open to use by the public was amusing.
There's this narrative on Twitter and elsewhere about how EU over regulated tech and fell behind US but this whole narrative is based on dismissing the Asian innovations. You have to omit China and other Asian countries from the graphs that show Europe is falling behind so to claim that free market and libertarianism is driving innovation and growth, because if you include Asia you end up claiming that its communism and centralized economies are actually the big success stories.
See, you have to realize that China is "still in the process of democratizing from free market exposure" and "the world is more peaceful now than it ever has been because of economic interdependence". Thanks Milton Friedman! /s
Admitting that offshoring was actually prioritizing lower prices (through regulatory arbitrage and human suffering based subsidies) over national economic and skills maintenance security will get you shooed out of the room. Yeah. Look at all the Indians lifted out of poverty... Shipbreaking multi-decade old ships, rebuilding lead batteries by hand (without PPE). Look at all the uplifting of the Global South! ...Often ruining their own environment to do things that absolutely would not fly here. Look at all the inroads we've made making the world more peaceful by shipping all the most advanced manufacturing infra over to a hostile regime that culturally values patience and tends to think further ahead than a 200 year old upstart.
There was a period of complete retardation in full swing in the 80's to early 10's that I truly wish it were easier to articulate in media res as opposed to in hindsight. But that's life for ya.
I think the general US ethos is something like “if we can offshore it, we will… find something where you can add more value”.
That said, I think a lot of non-US-based programmers grossly over-estimate the value of pure programming skills versus things like understanding a market, problem solving, empathy, English proficiency, similar time zone, professional trust, etc. on top of programming skills goes a long way.
No good programmer is afraid of their employment status, imho.
On the other hand, the number of offshoring nightmare stories abound, just like we did last time offshoring was the next hot thing.
> It has been quite some time to see some outstanding American innovation.
At a minimum, there is plenty of innovation that is not publicly known.
Is the US as prolific as some times in the past? Probably not.
Is that a fair comparison? Also probably not.
> Was it hubris? … because if you include Asia you end up claiming that its communism and centralized economies are actually the big success stories.
This sounds like sock puppet rhetoric.
As someone with a bit of experience in asia, including China, I’m happy to bet on Asia being good and sometimes the best at certain things, especially areas of incremental improvement.
That said, I’m willing to make large bets that Asian ascendency to the top of the general innovation leaderboard will not happen in my lifetime.
I agree, but IMHO when WFH many of these are weakened or even completely lost for the local American employees too. When WFH, the cultural integration(which provides the empathy, market understanding and other) gets reduced into whatever they can get from their neighborhood and from the internet. The internet part is available to everyone globally and the neighborhood culture is already not very strong in the USA apart from few places like NY. The rest are skills people can develop just as they can develop their programming.
It's true that good programmers don't have to be afraid but there're good programmers all over the word and when the market is not as hot, even the best will feel it.
Fair point, but that’s WFH done poorly, imho.
“100% remote” wfh teams should almost always have periodic get togethers on the company dime (among other things) to enhance/develop the relationships within the company.
Programmers will definitely lose some of their current market power in the future, but I don’t think that off-shoring will be seen as the main culprit.
I can’t tell you how much off-shoring is largely seen as a joke except by the VPs who are trying to get a promotion via cost-cutting. The quality of work needed is just not there at scale.
I'm not saying that the job of developers won't change or they won't be forced to be more efficient, but the work of turning abstract requirements into something a computer can execute isn't going away any time soon, even in countries with expensive labor.
https://equitablegrowth.org/aftermath-wage-collusion-silicon...
Of course the history of computer science is that we've had a continuous stream of productivity enhancing things that helped us become more productive. Whenever that happens, what doesn't happen is a decline in demand for programmers. As what we do becomes cheaper or easier, people want more of it. So far, there's little sign of demand for programmers declining. If anything, it's now so easy that lots of non programmers start getting their hands dirty as well. And why not? That just frees me up to do a bit higher level, more interesting stuff. A lot of programming work is tedious and repetitive. Getting some help with that is very welcome as far as I'm concerned. That just means I can build whatever I have in my head faster. So, it costs less for whomever is buying that. So, they might get back for some more; or raise the ambition level a bit. It's not a zero sum game.
The second thing that the article brought up (outsourcing) is an interesting one. Because they missed one important thing that LLMs also do: which is translating. Outsourcing programming work has been historically hard because of language barriers. Miscommunication due to people not speaking each other's language very well is a big challenge. You can make it work but it takes a bit of effort. I've worked with people that I can barely understand or that barely understand me. It's tedious and frustrating if you can't understand each other.
That language barrier is melting away as translating between languages gets easier. We're not quite at the babelfish level just yet but you can get pretty close if you have some patience. Meetings can get transcribed & subtitled (imperfectly) and those transcriptions can be translated. Anything in text form is easy to translate into whatever. In short, we still have cultural barriers but the language barriers are becoming less of a hurdle. The quality and speed of the translations keeps getting better. Real time conversation translations are basically no longer science fiction. And the quality of the translations is more than good enough for a lot of things now. Certainly better than two non native speakers trying to speak English with each other (I'm not a native speaker).
The value of hiring locally is mainly in that you can communicate with people directly. And that's just hard if you don't speak the same language. But that's not going to be as big of a deal as it used to be. The value of being local is still there but it's worth less.
The two added together means a vast increase in the number of programmers. But also in the number of customers. There are a lot of growth economies in the world. And they are all going to need programming work done. And they will want to hire the best they can afford and won't be hiring locally necessarily. We're not talking impoverished third world countries here either.
So, I'm not that worried. The opposite actually.