I think we should remember that AI is very young as a technology. Five years ago, it basically was unusable except in labs for proving that it could exist and now we can use it to do large amounts of simple programming. A couple of years ago, there was an AI-generated video of Will Smith eating spaghetti, which was hilarious. Now there is one which almost looks real. This rate of progress is phenomenal, and we can't really predict what will be happening in the next three to five years.
The trouble is that because of all the hype, everybody wants to use it for everything now. This means it's being used when it's not appropriate. It's being used for advanced tasks when it is best used for simple tasks and people are trying to use it where it simply isn't applicable at all. I realise the example above is to show how silly AI answers can be, but it also shows the kind of mentality that some managers have. Almost anybody can work out how to wash a car a hundred metres away; however, the manager demands you use AI and demands you follow the instructions it gives because AI is the future. That is the current biggest problem I think in using AI at work.
I do think AI will replace a lot of the work necessary for certain roles and maybe make certain roles no longer required. This is different to making people redundant. My role used to include spending bloody hours tarting up PowerPoint presentations to make sure everything lined up and used the right template. This part of my role is now redundant, as I can get an AI tool to give me a decent looking presentation, providing I put in the content and thought to give it the structure. It can do the pretty work.
I think that once the hype settles down, people will just look at AI and say, "Is it cheaper and does it provide a good enough result to work?" If it does, then they'll use it; if it doesn't, then they won't.
The reason I’m sure there will be a widespread effect on the economy is how AI is now being used in coding. Right now I’m doing more than 8 times the work than I could before (and fortunately have managed to get adjusted renumeration).
Right now companies are working on how to capitalise on these efficiency gains. Traditional automation of roles via software can be accelerated by using AI. It’s easy to point out holes in AI, but if a developer can use it to build more faster, this will accelerate.
I’ve worked with a Law firm who was already automating large elements of their business 5 years ago, they’re now moving to a significantly more comprehensive process with the use of A.I to create the code.
If you work at the DWP, then I bloody hope so. The simplest thing, like paying for your NI gap, requires endless hanging on the phone and then waiting months (literally) for a call back, They don't need AI, or LLM, or anything close. A Sinclair Spectrum would do the job better.
I heard something on the radio the other day when a politician or some other expert was saying about the youth unemployment figures. One of the big problems was that companies were becoming more efficient and more automated and better run, and that was causing unemployment for the young.
I'm not sure if he was implying that companies should be forced to be less efficient and less automated.
I know that in supermarkets in France, for example, the prices on the shelves are digitally controlled and automatically updated. We didn't bother in the UK because our workforce was so underpaid it was cheaper just to use people. I'm not sure this is a good or bad thing. AI will simply move this trend up the pay scales.
If I knew that then I'd be retraining to a job I'd be able to keep. However, at the moment, I don't think so.
Outside of the big american tech companies who are currently making loads of cash over this, are all other companies coming along for the ride? Has AI made those companies massively more productive? Growth says not. We've had the past few years of AI going to eat everything, it's not yet happened.
Will it? Who knows. We were going to have self driving cars 15 years ago, but they're still not quite there. I do think that LLMs are a fairly transformative technology, but I also think it's going to take longer for that to happen than all the very breathless articles out there.
The reason I’m sure there will be a widespread effect on the economy is how AI is now being used in coding. Right now I’m doing more than 8 times the work than I could before
As per some of the points and links above, that assumes that throughput of writing code is the major limiting factor.
I get your point in general though. I just don't think the economic effect will be directly proportional
https://pca.st/episode/cb1cb111-c970-4fdb-8ad1-5e4436f9400a
From the NYT.... Might be relevant.
Carpenter/joiner here. There are some aspects where AI, or rather technology, has had an affect on how we work, but it's mostly in CNC.
For what I do/did it's not relevant.
Same goes for many trades. Can't see AI hanging a door, fitting a kitchen, scribing round a wardrobe, laying pipe work, running cables, tiling a bathroom or trimming your bush.
Prefabs have been around for quite a while now, some very cheap, some very expensive. Haven't yet caught on in a big way, maybe someone in construction has a reason for that?
Simply put. Because of the cost.
Building a factory, infrastructure and the skills needed outweighs the current cost of having guys do the jobs onsite.
Had a project years ago around Stratford where all the wet rooms on site were craned in and bolted in place, fully finished. All I had to do was hang the door. Obviously the plumber and spark had to come along and connect everything, but otherwise each bathroom was finished.
As per some of the points and links above, that assumes that throughput of writing code is the major limiting factor.
For a standard junior/mid-level software dev jobs it is. Writing code, tests, associated documentation, and then going through the back and forth of - often over enthusiastic - code reviews is a boring slog which takes a significant amount of time. An average mid-level developer will do well to complete 2 or 3 features in a single two week sprint. Use AI (or LLMs for those offended by the term) dev tools though and you could potentially do that in a single day. And that's just mid-level or junior devs. For seniors multiply that again because done the right way you probably don't need the juniors to do the grunt work of writing the code and you can orchestrate agents to do it instead.
Once up to speed with the tools and patterns (which takes time admittedly) the time and cost savings are pretty mind-blowing. It might not apply everywhere but in software development that's probably 100s of thousands of relatively high paying (40-60k per year) jobs gone.
A quick chatgpt say there's approx 500k software development jobs in the UK paying an average of 50k resulting in 6bn per year in tax. Without even considering the multiplying effects that's a lot of tax income for the govt which is going to evaporate into thin air if they don't manage to replace those jobs with something else.
Who is going to hire those tradespersons
Your roof is still going to need replacing.
Your boiler is still going to bed repairing or replacing.
You're still going to need a new kitchen/bathroom/boiler/roof at some point.
Those basic maintenance jobs are always going to be there.
I'm some ways it may increase the quality of work in these fields as competition there increases. The main reason many people don't go into the trades is because of their parents expectations (more so if the parents are "academics") and not wanting their kids to get their hands dirty.
If you're using/relying on AI in @thols example then you should probably give up now and ask your doctor for a lethal dose of morphine.
Mate of mine is a lawyer for contract law or whatever you call it. Basically writing the same dross over and over again. His job is gone.
Got a few friends who work for SAP and they are worried about their jobs (ironically they're using AI to do their write code for them). Those jobs are potentially gone.
GF's sister sits there all day inputting numbers into excel tables making sure they tally up. Her job is gone.
Have a two friends who work directly for BMW (via subsidiaries) designing the presses that stamp out the panels for cars. Their jobs are already gone. One is now designing parts for drones used in Ukraine against Russia.
I've tried with copilot but I don't have enough knowledge to properly drill down into it.
Have you tried hitting it with a hammer?
Can't see my job becoming AI, although it can be useful. It's entirely about people, their experiences and interactions.
No knowledge of Copilot, so can't drill down and get into it? Kind of suspect irony there, but going with the flow: If only there was a readily available AI tool that could help you there.
Those basic maintenance jobs are always going to be there.
Did you miss the point? Who's going to pay for these maintenance jobs to be done? The economic risk here is a big down turn in employment, paired with continuing stagnation in wages, and more and more companies (and state bodies) around the world paying more to the US or China tech giants rather than on local employment. Money being diverted to the offshore tech giants rather than being spent on local wages means less money for trades people.
Did you miss the point? Who's going to pay for these maintenance jobs to be done
All the people whose jobs rely on face to face interaction and negotiation and are still employed, those dealing with complex physical tasks, those required to use imagination and artistry? And builders.
All the people whose jobs rely on face to face interaction and negotiation and are still employed
Why do you assume that? Replacing this is one of the stated aims of the tech companies pushing this stuff.
those dealing with complex physical tasks
Why do you assume that? They've got longer before they feel the impact, because despite the glossy videos the tech companies are a long way off developing the physical side of their "not as good as a human, but cheaper than a human" projects... but it's coming.
Anyway, assuming you're right, it still means that money will be drawn towards the tech giants abroad and away from the local market... which still means less money to pay trades people. The downstream impacts (positive and negative) of economic changes is often overlooked. If you want to picture it... when the mines were closed, it wasn't just the miners that were negatively affected, was it.
speaking of miners, are the (proverbial) ones who were told to 'learn to code' going to be joined back down t'pit with the former scrum master?
Who is going to hire those tradespersons
Your roof is still going to need replacing.
Your boiler is still going to bed repairing or replacing.
You're still going to need a new kitchen/bathroom/boiler/roof at some point.
Those basic maintenance jobs are always going to be there.
I'm some ways it may increase the quality of work in these fields as competition there increases. The main reason many people don't go into the trades is because of their parents expectations (more so if the parents are "academics") and not wanting their kids to get their hands dirty.
If you're using/relying on AI in @thols example then you should probably give up now and ask your doctor for a lethal dose of morphine.
Mate of mine is a lawyer for contract law or whatever you call it. Basically writing the same dross over and over again. His job is gone.
Got a few friends who work for SAP and they are worried about their jobs (ironically they're using AI to do their write code for them). Those jobs are potentially gone.
GF's sister sits there all day inputting numbers into excel tables making sure they tally up. Her job is gone.
Have a two friends who work directly for BMW (via subsidiaries) designing the presses that stamp out the panels for cars. Their jobs are already gone. One is now designing parts for drones used in Ukraine against Russia.
The numbers of tradespeople will increase in size as other job pools reduce, so the price for getting those jobs will reduce dramatically. Many people with less money and more time will do more DIY. I personally hire tradespeople because I’m time poor, when I go through periods of unemployment (I’m freelance), I do it all myself.
If fewer spend money on leisure, business for maintaining restaurants etc will vanish. It will affect everyone.
I am already replacing 4-8 junior staff as an experienced developer armed with AI. Those junior staff won't be buying houses when they're 30, so won't need them maintaining, they'll be in badly kept rental property. They won't be getting extensions built etc.
Have you tried hitting it with a hammer?
I just turned it off then turned it back on again
Money being diverted to the offshore tech giants rather than being spent on local wages means less money for trades people.
This is the crux of the problem. Politicians are going to have to get creative. The obvious one is a direct tax (or a tariff even) on the AI companies but we all know it won't happen for fairly obvious reasons. Which means we're about to see a huge chunk of the UK labour force outsourced to US (and maybe Chinese if they open them up) tech firms. The alternative is regulation of some form to discourage or restrict AI use to protect jobs.
Interesting counterpoint to all the mass-redundancy headlines, although seems to be a tiny bit conspiratorial regarding the motives of the researchers who are quitting jobs etc.
It is a bit emperor's new clothes. Everyone is trying to convince themselves it's fantastic. But rubbish in rubbish out, I watched a series of emails that were obviously run through copilot turn into nonsense and it didn't take long. It was so obvious that the protagonists were using a summarize and answer function.
A human is going to read a contract written and agreed through two sets of AI/LLM and find expensive mistakes and then the use for various tasks will be questioned.
AI wont replace me as I'm on the tools in construction. A robot may in so many years when they become affordable but I'll be retired then. If anything we can demand higher wages now than ever before. How long that lasts for is anyone's guess.
For all those that aren't too worried about their jobs because they are close to retirement you might want to remember that at least your state pension is not funded from anything you've paid in but from those still working so a reduction in jobs or just a move to lower paid jobs could still be a problem for you.
Only if you're relying on a state pension.
QC Inspector working for a company that overhauls aeroderivative gas turbines. I can't think of a single task I do that could be done by AI or a robot.
As with others I fear what the long term societal effects will be, but think that the problem is unstoppable.
Only if you're relying on a state pension.
A lot of people are, and even if you aren't actually needing it to get by it's still a significant part of most people's retirement plans
Of course it is but I'm replying for myself not others.
If AI reduces the resource required to produce a given piece of software by say 90% then shouldn't we see something like a 90% reduction in price?
Even allowing for some profiteering, the free market should dictate that bespoke software becomes relatively cheap.
The drop in price will create a big increase in demand.
So your team of 10 gets cut to 1 by AI. Only to be driven back up to 10 by the increased demand.
Seems that the mass job loss argument relies on demand remaining steady or even falling despite significant price drops, which I believe is illogical.
So your team of 10 gets cut to 1 by AI. Only to be driven back up to 10 by the increased demand.
AI isn't going to result in more software. Probably less if the recent crash in the share prices of SaaS companies is anything to go by. We're talking about the de facto disappearance of an entire professional discipline* in just a few years. That alone is going to be a major shock to the economy.
The economic multipliers are not more demand for software, but the removal of the physical and economic barriers of building it.
* yes there'll be others but looks like software development will be the first to fall
So your team of 10 gets cut to 1 by AI. Only to be driven back up to 10 by the increased demand.
AI isn't going to result in more software. Probably less if the recent crash in the share prices of SaaS companies is anything to go by. We're talking about the de facto disappearance of an entire professional discipline* in just a few years. That alone is going to be a major shock to the economy.
The economic multipliers are not more demand for software, but the removal of the physical and economic barriers of building it.
* yes there'll be others but looks like software development will be the first to fall
I guess we'll find out soon enough but colour me sceptical. I can't fathom why a significant decrease in the price of producing software wouldn't result in a big increase in demand. There are no end of possible applications for software which have hitherto been financially unviable. At the bare minimum you're still going to need someone who can take the client's requirements through to a finished product, even if AI does most of the grunt work, that's still a skilled job.
Of course it is but I'm replying for myself not others.
Well if you aren't going to miss it I can promise to see it's put to good use 🤣
I guess we'll find out soon enough but colour me sceptical. I can't fathom why a significant decrease in the price of producing software wouldn't result in a big increase in demand. There are no end of possible applications for software which have hitherto been financially unviable. At the bare minimum you're still going to need someone who can take the client's requirements through to a finished product, even if AI does most of the grunt work, that's still a skilled job.
This is my guess too. If you can cut the cost of developing a custom app from 100k to 10k, it will massively expand the demand.
The middle classes weren't arsed when it was coal miners and ship builders, maybe software engineers will be voting for right wing populists in 30 years time
This is my guess too. If you can cut the cost of developing a custom app from 100k to 10k, it will massively expand the demand.
Yes but that 10k is mostly being spent on tokens, not people. It may well increase demand for software development* but that won't result in more jobs for software engineers.
* Think it'll reduce the overrall amount of software. At the moment have all the user facing apps and systems that exist, and a massive amount of middleware and 3rd party libraries and services which they all depend on. AI removes the need for the middleware. Why use a library or 3rd party service when you can just generate your own in hours? That's why the SaaS industry is in crisis.
The middle classes weren't arsed when it was coal miners and ship builders, maybe software engineers will be voting for right wing populists in 30 years time
After years of indulging in smug potshots (thick, anti progress, peasant mentality, flat earthers) on the internet and on forums like STW. All from their safe space behind a computer. I am sure the red carpet will be laid out and they will be welcomed with open arms into the real world. You know the one they deny exists, except for other people who didn't try hard enough at school!
I don't relaly get what happens long term. Its trained on all code written up until now, but coding has evolved over time with new languages, design patterns, libraries etc. How does AI handle all that in the future? Do we just stop thinking about like we did with programming in machine code or assembler and let the AI take care of it? But the natural language interface is a lot less deterministic than a compiler.
I can't fathom why a significant decrease in the price of producing software wouldn't result in a big increase in demand.
A successful IT project is not just about writing some software. IMHO you need to think much broader and longer term than that.
Who is operating it? Who is maintaining it? Who is evolving it's features? Who is going to market it and monetize it?
Then...what if it was a really shit idea in the first place? Just like the many other shit ideas that are now flooding the market because it's cheaper to do the code bit.
I don't doubt that it'll be easier to churn more code out with fewer developers in the future, but that doesn't necessarily correlate directly to successful it projects.
How does AI handle all that in the future? Do we just stop thinking about like we did with programming in machine code or assembler and let the AI take care of it?
Yeah it gets very messy very fast. I have been using AI for development in a few different contexts over the last year or so since for my sins have been assigned to testing various of the latest and greatest AI dev tools.
For knocking together a greenfield site using well established patterns, languages and frameworks its okay? My specialism is backend work but it does okay there and as far as I can see okay at front end. It needs handholding and lots of "why is this your solution" and adding in rules saying "if it doesnt build try again". Its the sort of stuff I would knock together some high level guidelines and give to a junior dev or two to do before reviewing and asking "ermmm why?"
However I also tried using it to help me out with some stuff which had just been released and it was less than impressive to put it mildly. It proposed "use this really great method" which sounded ideal until a red squiggly line suggested it didnt actually exist.
On the flip side I have also been looking at some old code using a rather unpopular language. Again its crap since there no one has written some helpful notes that can be acquired by the model builder. Odd how in most cases their approach to intellectual property is flexible depending on if its their rights being violated or if it serves their interests
All that said the question is how different this is to outsourcing the junior devs. At some point the leads end up being too hands off to know the code and the people who do work for someone else.
I think one of the things which gives me a bit of faith about keeping a job is just how hard it is getting someone to turn business waffle into technical requirements and then into build notes.
The number of business analysts I have worked with is probably into triple figures but the number who could produce semi competent technical requirements could be counted on one hand even if I liked carpentry and was incompetent with power tools.
I've also seen comments on linkedin (used briefly out of necessity, honestly its become the worst social media platform IMO!) along the lines of:
"I've vibe coded an app, can someone check it for me?"
Which means can an experienced dev spend hours/days/weeks of time reading through 1000s of lines of previously unchecked code generated by someone who hasn't got a clue what they are doing..
I think there will be a lot of that in the future!
My opinion of AI and it's potential to go full Skynet or bring about the end of 'work' for the masses has changed a lot in the last few months.
When Microsoft presented to us what they thought the future was, it seemed bleak, they painted it as being very positive, but the gaps between the lines were massive. A lot of people who just sort of do 'stuff' were going to be made redundant, and I've seen it with my own eyes, people in the marketing depts of the big employers got put on notice.
What I've seen in the last 6 months or so is a massive walk-back, not a U-turn, but the estimated pace and scale of change was wildly exaggerated - what seemed impossible for economies to keep up with in terms of employment now seems more 'normal' more inline with changes we've experienced before. Jobs will chance, jobs will go, but those who can adapt will do well, and sadly those who won't or can't will be left behind.
I believe what we've experienced and continue to experience is an economic bubble, a bubble that's being inflated by those who are hoping to get out at the right time. Like the Internet, it will burst, there will be a crash and when it recovers it'll be more inline with what it's actually worth and then we'll know what it's going to mean in the real economy.
After years of indulging in smug potshots (thick, anti progress, peasant mentality, flat earthers) on the internet and on forums like STW. All from their safe space behind a computer. I am sure the red carpet will be laid out and they will be welcomed with open arms into the real world. You know the one they deny exists, except for other people who didn't try hard enough at school!
Not really sure who these comments are aimed at but when circa 6m people are unemployed and half of them never spend any money ever again and the other half all re-train from their laptop based knowledge jobs into the same narrow field of remaining "real-world" jobs, its going to be bad news for everyone if you ask me.
This is a good - but very long - read. Basically what happens to the banking industry and economy when mortgages that were based on white collar incomes can't be repaid?
https://www.citriniresearch.com/p/2028gic?utm_campaign=post&utm_medium=web
And these are the sort of ****s who run and own the AI industry..
https://twitter.com/TheChiefNerd/status/2025184575316471971?s=20
I'm pondering his use of the world "unfair" there... and no matter how I look at it, it's troubling.
its going to be bad news for everyone if you ask me
Of course more people competing for the shit jobs. Some will have to lower their standards across the board and might be surprised (certainly not used to) how badly people are treated in lower tiers of the employment pyramid.
Tools got faster, that is true, but companies still need people who understand what the tool is actually doing, where it breaks, how to design systems, and how to deal with messy real world requirements that are never clean prompts. Right now it is replacing some grunt work, not the people who can think through architecture, risk, scaling, and maintenance. The engineers who adapt and learn how to steer these tools usually become more valuable, not less.Software/Data engineer here. I reckon I've got about a year - maybe 2 max - before I'm redundant. Anyone else?
When ChatGPT arrived I used it create simple scripts and search for solutions to problems which I had to review or debug. Now with the arrival of Opus4.5 and 4.6 and codex 5.2 and 5.3 all I have to do is throw some rough requirements at it and it builds entire solutions which are pretty much perfect out of the box. Haven't written a line of code in over a month. PRs are pretty much pointless box-ticking exercises so hardly any reveiw or QA required.
The above was getting my slightly concerned until I saw one of my colleagues constructed a Ralph Wiggum implementation to build an entire app. 6 months of development done in 11 hours costing $1600 instead of $74k. Game over!
