GenAI comes for jobs once considered 'safe' from automation Jobs in geographical areas and scope once thought to be at low risk of automation are soon to be the most affected by generative AI, according to the Organization for Economic Co-operation and Development (OECD). "Generative AI will transform many jobs, but its impact will be greatest in regions that have been least exposed to …
Doing a lot more with AI? More verification of things that ought to be basic facts? More fact checking? More reformatting stuff because for some reason the AI did...
Expect your job to transmogrify into being a glorified secretary to a machine, where it's every inane utterance is your responsibility.
From the article: The latest report measured the impact of generative AI by how much of workers' tasks could become at least 50 percent faster through its use. ... The industries most exposed are education, ICT, and finance.
So US students, instead of graduating from school after the 12th grade, teachers will become so efficient that they will graduate after the 6th grade instead?
No, the purpose of school isn't to graduate the little human-resources, it's to keep them contained while the parents go to work.
Anyway, improvements in teaching efficiency won't shorten school it will just lead to the little geniuses leaving with 100 GCSE A**2**2 grades - or in American terms, a GPA close to Graham's number
The article doesn't really "say" anything! It makes a lot of generalized statements and generalized quotes to basically say that "maybe", "some people", will be affected by GenAI.
Hopefully, the people who get paid to produce this nonsense will get replaced by GenAI that can produce nonsense like this for 1/10 the cost!
You're right. Gen AI can never really do that. It needs the next step in AI to achieve that.
https://en.m.wikipedia.org/wiki/Artificial_intuition
Which has been in the works since 2017 and in use since 2021. Canada seems to be the hotspot.
There are four defined stages in AI. We see and use a cross between 2nd & 3rd gen.
The Equal Likelihood fallacy is the belief that when there are n possible outcomes, they must be equally likely and all have a probability of 1/n.
There is a related fallacy, which you have just demonstrated, which is the believe that if a process can be divided into n discrete steps, each one will take the same time and so the time required so far can be linearly extrapolated to conclusion. In reality the Pareto Principle fouls up this simplistic belief most of the time, and sometimes the later stages are simply impossible and the task is never achieved.
Producing a working fusion reactors is easily broken down: plasma confinement, ignition, surplus heat, power extraction. JET managed the first three over thirty years, so clearly a working fusion power station must only be ten years away. Right?
See also: self-driving cars, artificial intelligence.
This post has been deleted by its author
This post has been deleted by its author
Do I upvote because it makes sense, or downvote because the company won't care unless it hits the bottom line (in which case even more meatsacks will be terminated because "the computer never lies" etc)?
How about I do neither and instead offer a beer and think thank god I'm getting to be a crusty old git so soon enough this nonsense won't concern me...
The fiddling with numbers like 30% of tasks could be performed 50% faster makes my suspicious mind wonder if there is an element of ingenuous cherry picking. 30% of tasks? Is that by number or by time required to perform them? The 3 of the 10 tasks you perform daily might take 5 minutes which AI would reduce to 2'30". The tasks that take up 30% of your time might individually be performed in half the time but actually have dependencies on the 70% not amenable to the AI magick.
My guess is current crop of tossers running the shop envisage redefining all roles and employment generally in purely "AI friendly" terms so in their wettest of dreams if it cannot be done by AI it isn't proper valuable work.
Lovely world your grandchildren are likely going to have to attempt to exist in.
"so in their wettest of dreams if it cannot be done by AI it isn't proper valuable work"
Speaking of wet, my job in industrial scale washing up for a food factory. I would love to see a bot replace me. Because, you know, one needs to be highly responsive to the needs of production, to juggle multiple managers all of whom say their demand is the priority, manipulate objects of all sorts of shapes and sizes (including scaping off stuck-on cheese, nightmare!), plus of course the one working at the end of the machine has to not only arrange everything tidily, but perform running quality control on everything that comes out and know whether to send something back for rewashing or get the techs to come and kick the machine.
Yes, the washer machine. Epic with a capital E. Not the most reliable of things. I affectionately refer to it as "Mir" (as in the space station, because chewing gum and hope keep it going).
I would really enjoy seeing a bot come in, cope with that, and not break itself within the first quarter hour.....
I'm not sure I'd trust me with a fire axe, never mind the cow-orkers...
There is, however, a rather nice (huge) crowbar in the corner for when percussive maintenance is required. It's off limits as we're neither qualified nor paid to perform such actions. We just get to stand to the side and shout "harder! harder!" and see how long everybody can keep a straight face.
The old switch-a-roo, high productivity translates to a lot of former highly trained and skilled knowledge workers displaced from good paying jobs and now can only find work in low skill low pay service jobs or serving as low paid goons for the police state or worse as cannon fodder for use in looting weaker countries.
All those useless jerks in middle management positions will not have people to micromanage 24/7 Yes, 24/7. The 03:23 phone call from the AI saying that the system is down will still happen but it will only wake the boss up. With no humans left to do the job when AI falls flat on its experessionless face, he will be left to sort it out himself... Yeah right and pigs might fly.
Those of us left will be able to charge thousands a day to sort out their problems. At 2.5K per day, 30 days of work a year, would be enough for many to live comfortably.
....back in the real world
Phone call at 0323 wakes sleeping you in your assigned production facility dormitory, the AI ranks you on response time, time to resolve the issue, attitude, constantly records any 'malign' comments or expressions and notifies the dept for work and pensions system of your 'noncompliance' so that they can mark your record accordingly, cut off your 'allowance', apply punitive sanctions and notify the justice dept system of your 'malign' tendencies, lack of financial income and therefore the increased likelihood of criminal intent.
Should you attempt to forment any form of revolution/revolt/riot or even peaceful protest the security forces (meatsacks or androids) will 'terminate' your errant biological process
This post has been deleted by its author
Saw this story this morning:
https://www.businessinsider.com/middle-manager-hiring-white-collar-recession-layoffs-jobs-efficiency-2024-12
When those middle managers are booted out of the company, who's the AI going to call? Someone in the C suites? That ought to be entertaining.
The answer is "working managers". My employer has worked this way for years! Our managers actually do the work, along with the people they supervise. Which is very lean!
It also requires that the people they manage not require "micro-management", that they can do the job, get things done with little to now active supervision.
If this turns out to be a reality, it just means that the nature of the affected jobs will change -- from spending time making informed decisions to instead wasting it working out whether the AI is talking bollocks.
Upvoted because I am literally seeing this happening now!
Hours wasted trying to work out why what is being presented does not reflect reality, as opposed to dealing with the problems we are aware of.
Take skilled well paid position (i.e. skilled trades) and turn them into low skill McJobs - i.e. the replacement of machinists with CNC "loaders" whose only required skill is removing previous work piece, cleaning out anything that might foul, inserting new blank,closing door and pressing "go" and all for minimum wage, no perks/"benefits", no job security, no sense of pride or prospect for advancement, just utter and total monotony and job insecurity.
Just as Adam Smith envisioned in "wealth of nations" except he saw it as a good thing while missing the risk of a rise in poverty due to it
I wish my job had been 'downskilled' I could spend the day in bed instead of wrestling with a downed robot that had a wonky sensor, by the way the 'CNC loaders' as you describe them are also out of a job due to the aforementioned robot. and its brothers.
We have 4 unskilled left now.. and 2 apprentices. everyone else can do skilled jobs or highly skilled jobs, or do the complete job from planning to programming to setting the robots up.
Oh and everyone misses the bit in Adam Smiths "wealth of nations" where he says its important to reward managers, investors and labour for their efforts, without one of these, every company falls over and dies.
Hint: productivity and wages rose together from 1945 to 1980, since then wages have matched inflation while productivity has risen by another 75%. somethings going to give.....
@jospanner
Did the white collar genuinely believe they were immune?
Yes, because as office workers, they always have and always will think they are superior to the shop floor workers. How about that insulting "dress down Fridays" for an example. As in let's dress down so we can pretend to be blue collar workers.... Twats
Well, once upon a time... in a hugely big tech co (let's call it HAL for now) HQ not so far away in the north east USA, ... where the upstate managers used to play... there was the old joke....
Three lions escaped from a local zoo, they went their own way but agreed to meet up 3 weeks later.... When they meet up 2 of the lions are looking thin, stressed and mangey, but the third looks well fed and happy.... "what happened?" says the third, the first replies "it was awful, I couldn't go out I had to hide all the time, there was nothing to eat", the second replies "they came after me with guns and chased me, I had to hide in bushes, I was terrified, but what about you?... you look great!". The first replies "Oh, I hid out in the HAL car park, I ate one middle manager every day, and nobody noticed...".
Sometime later the HAL boss came along and fired all the surviving middle managers over 40 and fed them to the lions....
The AI Lion is coming... it will eat all the remaining middle managers... no one will notice....
Oh FFS, based on initial office comments - young whippersnappers!: Arthur C Clarke - 2001 A Space Odyssey! You saw it once and didn't understand the ending... H...A...L... 9000 - move each character one to the right in the alphabet! Arthur did have a sense of humour. <sigh>
This post has been deleted by its author
When AI replaces all the bean counters and share-traders then we may have some return to having real intelligence running companies again as the bean counters are basically historians who live in the past without any ability to foresee the future, Creative accounting is a no no.
Visionaries are required as managers of organisations not historians. Visionaries are often amiss with timelines and often have to be checked, but without them we would still hitting each other over the head with stones.
Real intelligence is what is required not artificial.
I suppose I will have to start working on improving the training data.
I dare say the pipeline currently would be something like:
Get a pre-trained LLM
Apply local knowledge to it from a local training dataset
Attempt to use the result for a local task
Identify results that are sub-optimal
Generate another model that identifies elements of the local training dataset that most likely contributed to the sub-optimal result
Review those local dataset elements and improve as necessary.
Retrain the pre-trained LLM on the improved local Dataset
Wash-Rinse-Repeat
If we have a model that can do this then no knowledge workers need ever to get out of bed again. Just chain them together for any arbitrary degree of utility.
I do not think it is likely that we will ever have such a thing so I will always have a reason to get out of bed of a morning and go to work. A pity really. But at least I will still have a job.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
