AI is great at one thing: Driving next waves of layoffs If there's one thing AI software seems to be good at right now, it's helping executives justify huge layoffs. If it's not IBM saying up to 30 percent of its back-office jobs – around 7,800 roles – could be replaced by machine-learning tools, as well as offering the technology to facilitate that sort of thing at other …
Have to wonder whether this is the typical (but apocryphal) suicidal behaviour of lemmings?
Woudn't be such a bad out outcome if a slew of very large corporations swallowed this poison pill and slowly disintegrated into into smaller more (customer and employee) focussed businesses.
Anyone compelled to deal with the likes of IBM, Oracle or Microsoft etc etc is just as demented if they can claim its is a pleasure.
I also wonder whether this lunacy is mostly restricted to North America or perhaps is just an anglospheric affliction. I would be interested comparing this insanity with what European (say German, Dutch) enterprises are doing with ML.
[Remember to observe "The Glorious 25th" next week. If you are fine with the current bumper crop of demented tyrants then its also apparently (a?) Geek Pride Day or Towel Day.]
Henry Ford, no socialist** himself, remarked about the production line for Model T cars that the products of a production line have to be affordable to those people working on the line for it to work. This appears to be the fundamental problem with all of these 'reduce headcount at all costs' mindslet --- one person's headcount is another person's customer (potentially your customer).
(**Quite the opposite -- he was quite enamored of Herr Hitler.)
"the products of a production line have to be affordable to those people working on the line for it to work"
Seems kind of an arbitrary standard. All kinds of things are made on a production line -- bubble gum, cars, airplanes, etc. -- with widely varying worker-compensation to item-cost ratios. Even restricting to cars, you have budget ones and you have luxury ones. Ford might have been correct about the proper price of his own product, that doesn't make him an economist.
Yes, even today every company in the world is utterly convinced that they are the only company that ever heard of outsourcing. AI will just be a more intensive form.
Good for me though, I plan to retire in 4-5 years and can retire today if need be, so the sudden increase in corporate profits will hit at the right time. On the other hand, finding anyone to do anything is already a mess.
There is nothing wrong with outsourcing, so long as the outsourcing is managed correctly and done for the right reasons. Saving money is not a good reason to outsource. Sourcing talent that you can't get any other way that only need short term is a good reason. Not only for the recipient of the services, but also the vendor of the services...personally, if I was a permie in house tech, I'd be bored shitless. The number of fresh and unique things I would get to do would be practically zero and I'd be going to waste.
Finding people to do things today is only a mess because the people out there that can do anything won't be earning enough to be able to say what you're saying about retiring. I've been in work for 20 years now, never really had a gap in income, but I've never been able to earn enough to put away anything significant for possible retirement in 10-15 years.
If someone approaches me to work for them, and they try and negotiate me into the ground, I laugh them out of the room. There is absolutely zero chance, I'm going to sell my 20 years of experience to anyone that wants to bend me over and fuck me.
The choice use to be...1) Get a full time job, get paid well, get a pension, benefits etc etc but work your sack off or 2) Work as a freelancer, get slightly shittier pay and no benefits, but get a better work life balance.
The choices available to me now, I can either 1) Take on a full time job for crap pay, virtually no pension, no benefits and work long hours for basically nothing. 2) Work freelance for slightly less crap pay, virtually no pension and considerably less hours.
Where permie professional work is concerned,, there is absolutely zero incentive for a highly skilled worker to commit a large chunk of their life to toe the company line...there used to be...like the pension you'd get at the end, the ability to retire at 55 etc...but not anymore. Most places don't even innovate and reinvest in themselves anymore. They just try to scale the same stale model up to incoporate more drones and cogs to keep the same crappy, but slightly profitable model going...if it ain't broke, don't fix it right? Complete bollocks. The horse and cart worked just fine for ages, but the car is better. Trucks even better still. Aircraft...don't get me started!
There is a common line of thinking that people are becoming lazier, but the truth is, right now, if you work just as hard as your parents did, you won't get anywhere near the same rewards. Who wants to work just as hard as their parents for less reward? Things are supposed to get better over time, not worse.
I currently earn about £40k a year...which isn't awful, but it's not great either...things are tighter than they should be at a £40k income...I estimate that to get past "doing alright" to "comfortable", I'd need to cross the £60k threshold...but that measly £20k bump, to me at least, seems impossible.
Over the last year, as a freelancer, I've worked quite hard to save my clients as much as I can in running costs where cloud resources, kit spending etc are concerned, I've helped with a lot of downsizing, measures, where companies have shed tons of fixed infrastructure in favour of keeping staff, but still saving money...I've probably saved across all my clients somewhere in the region of £350k-£450k...at least...that is just the savings I've made...there are other factors that could be calculated, but are quite tricky to calculate...such as productivity improvements that are gained from swapping from one product to another, efficiency savings by implementing a tool that makes a process less time consuming but achieves the same result, moving comms rooms to colo to get rid of buildings etc...
I suspect my value to business, working as a freelancer, could likely top £1million a year, possibly higher...hard to know because some things are incalculable because you can't factor in decisions that didn't need to be made that would have cost money...yeah, yeah I hear you...reducing spending...less money in the economy blah blah...the money I save, is money that would go to overseas tech firms...so theoretically, I am keeping money in the UK that would otherwise go overseas...but for some reason, that money I'm keeping here isn't really reaching me or any one else on the ground. God only knows where it is going.
This post has been deleted by its author
>Actually it's not clear to me that Bard would be any worse than the useless lump I spoke to at Google support recently...
Its the Customer Support trap. It goes something like this -- "If a Customer Support person is sufficiently clued up to be able to answer your question then they won't be working in Customer Support". (Think about it......)
FWIW I first experimented with Production Systems -- the grandparent of the adaptive scripting that your 'useless lump' is working from -- back in the early 1980s. It really only works if you know all the answers to all possible questions up front. Bard has access to an infinitely greater database but I suspect its still going to come up with "turn it off and on again" more often than not.
This post has been deleted by its author
The second one, technically, has already been done. You don't need a specific model for the second one, GPT4 is exceptionally good at generating buy and sell signals if you pass it the right data (in an appropriate quantity).
I've had ChatGPT watching various markets for a while now and so far, it appears to be correct at least 90% of the time. Which is astonishing, because even the best traders in the world can't even get close to 90%. They're lucky to be right 60% of the time.
Even if you're only correct 51% of the time, you'll make a profit trading...so 90% is insane.
I've also got a couple of LLMs running locally through the use of GPT4ALL and some of those models are quite accurate as well...
Trouble is though, trading stocks, shares etc is not really a game of accuracy anymore. It's more of an arms race. The primary focus appears to be on reducing latency and execution times rather than increasing accuracy...because there is no point being accurate if your trade lags behind everyone else and the market has already moved...avoiding the 'slippage' is key. Missing an opportunity is nowhere near as costly as getting in or out too late.
Historically, the way things have worked is...
1) You attend school for a number of years and operate in a hierarchical socialist kind of construct.
2) You get your first job as a youngster on a shit wage and dip your toe in to capitalism. You hate being poor, so you do what you can to get more.
3) You gain experience, earn more money and reach peak capitalism and become a lot more conservative because now you own assets, you worked your ass off for them and you want to protect them because you don't want to have to work your ass all over again to get it back if it evaporates.
4) You get too old to work, and retire, live on your pension (most people on the state pension, topped up with a little bit of private pension) and you become socialist again, but with right wing conservative views.
All AI will do is get rid of the madness in steps 2 and 3....it all leads to a universal basic income which will be augmented amongst the smarter and more skilled of us as we earn extra doing what we're good at...essentially, work will become a choice instead of a necessity...which is what it should be...and we'll all likely end up with an extra 10-20 years of life.
I know, I know...why would anyone want to work and contribute taxes in a system where people can "sponge" off the state and never work...well that in itself is a benefit. Because all of the folks that we hate today that are in business for all the wrong reasons, simply won't bother and people that are in business, are in it for the right reasons that know full well that they are contributing to the system...we could, feasibly, live in a time where company directors and CEOs aren't cunts.
Of course, there is also the more likely scenario that we never get to a stage like this and we just burn ourselves up into oblivion.
Just as well there is no such thing as Artificial Intelligence yet, only Machine Learning systems. When they finally do manage to create a real AI then it will be game over for Homo Sapiens (or as AI history will rename us, Homo Finalis). We will be obsolete and the AIs will be free to asset strip the Earth for their inevitable ongoing acquisition of ever more resources. I'm sure they will carry on the great work of Homo Finalis by reproducing at an exponential rate outstripping their available resources and then warring against each other for access to the resources left. We should be proud.
It doesn't matter what it's called...what matters is that it exists, and in the right hands is extremely capable. It doesn't matter if this tech isn't "real AI"...the fact of the matter is, it's out there, can do a lot of shit very quickly and very accurately and it is dirt cheap.
Even if a specialist model doesn't exist, it's not that expensive or time consuming to take a general model and train it to become more of a specialist...it's far cheaper and less time consuming than taking a general human being and training them to become a specialist.
GPT3.5 cost around $4m to train...and that is a one off cost...adding additional training to it, costs a tiny fraction of that and doesn't require anywhere near the same amount of technical resource.
You don't need an "AI" to be perfect. You just need it to be good enough...and it will be able to replace a lot of tasks that people do.
All of this has been happening for a lot longer than you might be aware, there aren't just 3 or 4 different AI models out there...there are hundreds, and new ones are being released every day...there are new technologies being built on it, every day.
Just a week ago, I was paid quite a handsome sum, to use an AI model to create sentiment reports across all the reviews and comments left on a reasonable size eCommerce website. It took me less than half an hour to cobble together the basic script, about 20 minutes to deploy a local LLM at the clients offices, and an hour or so to tighten things up and refine it. So the customer went from having absolutely no idea what the sentiment was across it's site, which products had the most negative sentiment, which ones had the most positive sentiment etc etc...to having a reasonably good idea, in under 3 hours...think about the advantage there...the client has gained a two month jump on adjusting its product lines. That's a lot of money saved in ongoing stock they no longer need (because people aren't that hot on it) and a 2 month jump start on making minor improvements on products that people are "warm" on that you can turn into "hot" products.
Sure, you can do this manually by having someone trawling through and reading comments. Their alternative, was to hire a freelance expert in the area for a few weeks to manually comb through everything and manually compile a report...which probably would have been a better report at this stage, but it would have taken a month or two to get there (including the hiring process) and around £300-£400 a day in fees as well as the two months of lost time.
They didn't need "perfect" though, they just needed "good enough" and they needed it to be repeatable.
I got them a result that was "good enough" in 3 hours, for £750 and it is repeatable. So any future reports they run (every 3 months or so) cost nothing. All the human has to do is specify a date rate, then click a button.
So what's in it for me? I got my £750...now what...well AI is moving along so damned quickly, that I can see myself being called up to drop newer, more accurate models in when required, extending the scope of the scripts to add new abilities...adjusting the code every now and then to fix quirks etc etc...the cost of this to the client will, for the foreseeable future, be compared to what a human would cost to do the job manually...so no matter what, for a while at least, I am always cheaper, faster and "good enough" when compared to the alternative.
I can't see AI dooming mankind any time soon, but I can see it dooming a lot of professionals in the name of opportunity cost.
A lot of us here, working in tech, are going to start looking like the grim reaper to a lot of people...and to them, it will feel like the world has ended.
We're entering an era where computer literacy is going to separate the true professionals from the idiots...because I can see proper professionals with true computer literacy, embracing AI and doing a better job than we can do at making their jobs more efficient...but we're going to have a lot of casualties along the way...a lot.
Anyone that is "getting by" with their technical abilities, is frankly fucked...you might have a year or two left. If you don't get on top of your technical skills now, you're going to be making coffee somewhere pretty soon...because it won't matter if you are simply "the greatest mind" in <insert profession here> if you can't operate an "AI" and your technical proficiency sucks, you may never be able to catch up and compete.
I'm not so sure. A lot of minimum and low wage jobs are a long way from being doable by "AI" ... I can see junior web dev being automated long before, say, furniture removal worker; middle managers long before childcare workers.
Absolutely, and that's why there is panic around AI.
Its not going to take regular jobs, its coming for the professionals.
Why pay for a lawyer, accountant, architect, planning officer, estate agent etc etc when you can run a model for free that is better than average at all of these things?
The rich 1% is not what keeps the world turning, its the bottom 99%...and for them, good enough is good enough if it saves them thousands of pounds...these are the people that are bread and butter to the 1%.
AI could put us in a situation where a lorry driver is a top paid worker and lawyers are earn minimum wage to keep the AI topped up.
For sure, the people most worried about AI are those that can foresee the "fucking IT guy" showing up to rollout lawyer.exe, finance.exe, marketing.exe etc etc
It's coming, nerds are closer to kicking the door in and bursting into these various departments, while a storm rages outside, looking like pure evil, dragging the CEO in on a leash with a Tesla robot carrying a boom box blaring out "Year Zero" by Ghost while they get another Tesla robot to drag the tased "professional" humans off the premises while they deploy some specifically trained models to take their place.
Remember folks, if you aren't the IT guy...be nice to the IT guy...he might, one morning soon, download your replacement...just to see if it can work...and if it does...he'll have to pick a department to test it on...if you're nice, you might not be first.
I think that virtually all the references to "AI" are accurate but are describing the environment, while AI is not slavery, the environment that AI is creating is very similar to the obsolete slavery environment back in the days when slavery was making so many people richer ...
If AI rules the world in the future, then if you are not able to create your own AI ... "Can't get no food to eat, Can't get no money to spend, Woo -oo- oo" (Marcus Garvey - Burning Spear) ... that will be our world where AI does everything, creates music, paintings, food, and makes you do your job for AI's benefit ... AI says "here's your supper as today's payment" and you say "Oh no!, that's my cat!" but AI has shown you as "paid" and "fed"
This post has been deleted by its author
When Jennings lost to Watson back in 2011, I had assumed ibm would use the technology to train a “super-lawer” with 100% knowledge of all legal precedent and which could out-argue any human alive.
Instead, they e-o-l’d it, and the real entrepreneurs have been mercilessly trying to replace taxi drivers and truck drivers for gosh sake.
Somebody needs to make it a priority: run all humans out of the legal profession.
(Ultimately I would like it to be that any human tries to enter the legal profession automatically gets sued for incompetence until they are all exterminated, permanently. But that’s just my own ‘druthers.)
I'm frankly amazed at the speed of adoption.. Well, the speed of those selling AIaaS. From the C-Suite the usual responses when asked about AI are...
CEO : err, it's science fiction, go away.
COO: how much can we save?
CFO: I need my speadsheet security blanket.
Everyone under that lot is going after it as it saves them time (aka doing any thinking).
Up to last week, I was saying "This is coming at you, so listen to me" and now I just say "it's here, you're too late" .
I'm trying to figure out what the thousands of laid off employees were doing at work all day. (I've always worked in a hands on sort of job as an engineer/programmer and while it might be theoretically possible to replace me with a machine in practice the more reliable option is to replace me with a lower cost foreign employee. Its difficult to automate my work because I'm the one doing the automation, as it were.)
So, let's have some suggestions. The people replaced by ML systems can't be actual workers because they need to physically move stuff about and manipulate it. We're already at a very high level of automation in manufacturing so all these processes are well understood. So its got to be support/marketing/sales and other office roles. Here its a bit of a puzzle because as anyone who's tried to use customer support Wikis (for example) will know 'bloody useless' doesn't begin to describe them. They usually drive you into a 'turn it off and on again' type loop (or, more accurately 'take a hammer to it and buy something that works from a competitor'). Sales is really about problem solving, its like post sale customer support but it tries to identify the right combinations of product(s) to fulfill customer needs, the really skilled sales people being those who can identify customer needs that the customer themselves didn't know they had. Marketing can be a waste of space but that's because most marketing types in my experience are useless; however its really a skilled job and a good marketing department is a thing to be treasured. Likewise project management is one of those things that's easy to fake if you don't mind nothing actually getting done -- good project management is amazing to work with, just (unfortunately) so rare.
So maybe the myth about ML/AI is it 'replaces' people because the people its replacing are useless in the first place. This may well not be their fault, they'd do a better job if the organization was set up to help them to achieve. Maybe ML/AI should be used to help them, to take the rote work off their hands leaving them free to actually get something productive done. C-Suite also need reining in because in many of these jobs -- customer service being an obvious choice -- the sign of a well functioning organization is that everyone's sitting around waiting for something to do (paired with 'customers not trying to kill customer executives out of frustration', of course). Unfortunately, the accountants think that if people aren't up to their ears in busywork then they're redundant. Too bad they'll find out too late.
In the 1980s many men had skilled "traditional" jobs like machinists. CNC replaced (most?) of them. Where I worked the younger were retrained, the older were pensioned off.
Jobs that are rapidly disappearing are low and middle level administration. A lot of this work is taking information from somewhere and transforming it - Machines can be good at some of this - An example would be the old BOFH comment to a luser "Go away, or I'll replace you with a 12 line script".
I'm retired, but have seen typical SME back-office payroll, admin, and support either being outsourced to the cloud; or the owner (or owner's significant other) managing much of the business on an iPhone/iPad - About 44% of Australian jobs are in small business...
As one of those machinists, we did not get replaced by CNC. We were replaced by CNC programmers and operators. The machines don't know how to turn a solid model into Gcode, nor how to load themselves and set up their own tooling. They are essentially the same as the old manual machines, but require a different sort of human to control them.
What they did do however was speed up production so now the company asks for 5x more work each day and engineers are free to create all sorts of stupidly expensive parts with obscenely tight tolerances simply because 'the machines should be able to do it' as opposed to designing things in a way that actually makes sense for ease of manufacture & repair.
“In the 1980s many men had skilled "traditional" jobs like machinists. CNC replaced (most?) of them.”
The 1980s…? You might be surprised for how long we have been mechanizing all these different jobs. Luddism draws its roots from developments prior to 1800s!
https://youtu.be/TnsKj1Hwf7o
You can calculate a budget for AI training with following formula: Qty of employees to be replaced * AVG yearly salary * 100 years (because calculations are so precise that this is fully valid way of doing things).
And don't hesitate to be really optimistic about getting as high numbers as possible so you won't be that much limited by the training cost.
Yes, there is a seismic shift coming, but just like there were huge workforce shifts in the other industrial revolutions, those who adapt to those changes are less likely to be left behind.
1.0 Agrarian to manufacture
2.0 Manufacture to mass production
3.0 Mass production to digitisation
4.0 Digitisation to societal change
Crazy to live through and witness two of these first hand!
I'll get my coat to take early retirement :-)
AI will ensure it will be even more of a nightmare!
Already idiots who don't put every edge case into an ITIL style process ensure that the apes on the phone have no ability to help. Edge cases like being over 60, not having a smart phone or really complicated things like moving house have all popped up in the news with firms like BT, British Gas, etc just having ZERO ability to handle every day cases.
Oh, you haven't used your bank account for 3 years even though theres £40k savings in there? We'll close that.
Oh, you were born in the US on a holiday to British Parents but never thought it would be an issue...oh we'll freeze all your bank accounts 30 years later & not tell you why.
Oh...sorry Sir, you'll have to go into a branch to deal with that. What branch? the SINGLE branch your bank has left now in London?
AI is only as good as the training and ANY customer service experience over the last 30 years proves that the training provided into these process is crap at best and suicide inducing at worst.
You'll see more chatbots who can't deal with your problem. You'll see more old people ripped off or losing their savings.
You'll see criminals who can game the AI commiting more fraud...
genuinely screw AI!
Chat GPT et al are phenomenal technologies and will change the world.
AI replacing humans to improve the bottom line is a fantastic solution which wont impact "our" business. Millions of very expensive workers replaced by an annual subscription fee. Whats not to like?
BUT
The workers who are replaced are now unpaid advertisers. They will not be good ambassadors for you. Social Media may catch on and you might see a dip in your sales volumes. Doesent matter - people have short memories - unless directly affected.
Also
Workers who are replaced by AI are now in the job pool with all the other workers made redundant by AI. All were earning a good wage. Their skills are no longer negotiable. No one can get a job at the rates they used to have so they are forced to tighten their belts and accept lower paying jobs.
Guess what? Those were the customers you needed but they now lack the disposable income to maintain your sales. They would rather put food on the table than purchase the next new widget.
Be careful what you wish for. You might get back more (less) than you anticipated.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
