Basically, any white-collar job (management, technical, administrative, whatever) is under threat.
Mo Gawdat, the former chief business officer of Google X, has a stark message for white-collar professionals: Artificial intelligence isn't just coming for entry-level work — it's coming for everyone, including software developers, CEOs, and podcasters.
In a Monday conversation on the "Diary of a CEO" podcast, Gawdat predicted that most knowledge workers would be replaced in the next decade and said many still underestimated just how rapidly this transformation would unfold.
He cited his own startup, Emma.love, which builds emotional and relationship-focused artificial intelligence and is run by just three people.
"That startup would have been 350 developers in the past," he said.
. . .
But he warned that AI was being deployed by people and institutions driven by profit and ego, not ethics.
"Unless you're in the top 0.1%, you're a peasant," Gawdat said. "There is no middle class."
He predicted a "short-term dystopia" beginning around 2027, driven by mass unemployment, social unrest, and an economic structure that fails to adapt.
There's more at the link.
We're already seeing this in operation in many knowledge-driven fields such as finance, real estate, etc. A professional is now expected to use AI to augment or supplement his training and experience, conducting searches, market research, etc. in the background while he applies himself to current problems in the foreground. It's reported that productivity improvements of up to several hundred per cent are being claimed - and those who aren't "getting with the program", learning to use AI to work smarter, are already finding their careers being sidelined or cut short.
It's not just in America, either.
The latest Office for National Statistics data shows that job vacancies in the UK have fallen for 36 consecutive months, obliterating the previous record of 16, which was the result of the global financial crisis in 2008.
At the same time, the past three years has seen the rate of redundancies almost double from 55,000 people per month to 114,000, and the impact has fallen disproportionately on middle-aged and older workers. More than one million people have been made redundant over the past three years, with 34 per cent of the job losses hitting people between 35-49, while 30 per cent were aged 50-plus.
What is truly frightening is that all of this has occurred during a period of economic growth, albeit weak growth. The general consensus is that when an economy grows, jobs are added, and when it shrinks, jobs are lost. Yet this relationship appears to have broken down and experts believe this is down to a convergence of factors, including the rapid advance of AI.
Again, more at the link.
The AI conundrum is of particular interest to writers, people like myself. There are already literally thousands of books self-published on Amazon that were "written" entirely by artificial intelligence software. One AI program can be given broad outlines, then produce a novel set in dozens of different genres, using different character names, settings, etc., but all basically the same book. It can generate them in a matter of a few days. Amazon tries hard to intercept such books and stop them from being dumped on the market, but an acquaintance there tells me it's getting harder and harder to detect them. Early AI "composers" (for want of a better word) were amateurish, couldn't parse grammar very well, and were relatively easy to spot. Newer software is much more capable, and it learns from its mistakes, becoming even better with every iteration. I won't use such software - I take pride in producing original work, thank you very much - but many people out there have no such scruples.
Food for thought.
Peter
24 comments:
Category Romances were already basically the same plot, same characters, different costumes, and they were the Romances written for the Romance addicts. (I worked in a used bookstore that specialized in Romance, we had three ring binders with ‘if you like this you might like this’ info for the patrons)
One important facet of work is that is a communal activity. When I work I take care of the people around me. I'm retired and worked for a bit at a big box store. How they tried to sap any purpose or reward out of helping your neighbors. AI is following the trend. As a manager, I had to learn my people and what motivated them. Do you want an AI bot to motivate you? Monitor you? Decide you talked to a lonely widow customer too long?
AI is coming, but it does no better than it is trained. I think an organization that puts its people and community first will remain competitive. Companies now advertise they are green, what happens when they advertise they are silicon free?
AI could easily replace all teaching positions from pre-school through graduate school but what, in most cases, would be the point in educating people?
I'm not as pessimistic about AI taking jobs. I mean yes it going to take some of course but it is worth noting that the current mass investment in AI datacenters seems to be burning enormous sums of money for very little revenue. It is not at all impossible that the AI bubble will pop as the world discovers that people are not willing to pay for AI at a rate that allows payback on that investment because that cost is the same as, or greater, than the cost of hiring humans
I did some calculations a few months ago - https://ombreolivier.substack.com/p/ai-actively-incinerating-cash?r=7yrqz -and nothing I've seen since has shown me that I'm wrong
I've seen many new engineers (real engineers, not the software geeks) get completely locked up because there's not always a clear and correct answer for the problems you find in the field. The textbooks from school all had neat and tidy problems, the real world has messy weird overlapping problems. I come from a long and distinguished line of "I don't know, let's try it and see what happens" style of engineering. I really don't see AI replacing that. But then again, I don't see many new engineers replacing that either. Once I retire (soon), they'll have to either get messy or give up.
The book thing is new to me, but I guess I'm not reading high literature anyway. I'll have to look back at some of the trashy sci-fi / fantasy stuff I've read recently to see if there's anything weird there. I can't see AI getting humor correct, but what do I know?
AI is a computer program. As all programs, it is biased in the original coding and is bound by the law of garbage in garbage out.
Yes, AI programs check all over the web and then produces an answer. The problem is that the answer is just an answer, not a true, correct, verifiable answer.
So, we are pushing toward making business actions even more ignorant than in the past.
Dave
We already know that AI makes stuff up (as several lawyers have learned to their chagrin) and it discriminates in hiring (as Workday is finding out). 9 times out of 10, the AI summary at the top of any Google search I do for work is incorrect. I recall reading an article by a data scientist where a company had the same data analyzed by a team of people and by an AI, and within a few months, the AI data was so far off it was basically fiction.
AI is only as good as the prompt and the data it has access to. will it get better? Yes, eventually. In the meantime, companies relying on AI over human intelligence are going to face tough times.
Call me when AI gets advanced enough for self-driving trains, which have only three settings: forward, stop, and backwards.
We won't even talk about more complicated functions.
AI will only replace actual retards with virtual retards, on its best day.
Period.
If you have a make-work job, you're in jeopardy.
If your work product is hash in the first place, AI's got your number.
And that's it.
AI can't bake a cake, change a tire, let alone think an original thought. It won't write Romeo and Juliet, War and Peace, nor Beethoven's Sixth Symphony, nor anything like.
Skills will remain untouchable.
And humans using it will break the Three Laws Of Robotics with it before breakfast.
It will, in short order, become the mark of the crook and con-man, and be about as career-ending as being found with a running iron in your saddlebag in cattle country circa 1880: one order of rope necktie in 3, 2, ...
Self-driving trains are common in large subway systems, and were technically feasible from about 30 years ago. There were still driving cabins for emergency operation, but some new Metro lines already use trains without driving cabins.
Note that is not AI, but relatively simple engineering, especially advances in signaling systems.
Nah, what will happen is that the business owner will get flim-flammed into firing his engineers to replace them with AI.
Then he goes out of business when the AI gets the code wrong, and the installation process puts it into production.
Replacing your engineers with AI is like replacing them with interns because they're cheaper. Some companies do that; but their track record isn't good.
I expect the result of massively increased AI in the workplace will be a mixture of tragedy and farce. Entry level jobs will be mostly obliterated. My first job, as mailroom boy for Miracle-Gro plant food, could be almost entirely automated with ease (except for me cleaning the bathrooms and emptying the waste baskets.) The Gen Z Stare will be replaced by a kiosk that doesn't condescend. Middle management will become an endangered species, since HR, payroll and many other tasks could probably be better performed by a machine. The farce will be that some zealous fools will try to convince us that touch screens can hear our confessions, teach our children and resolve our need for meaning. I just hope I pass before it gets too stupid.
For a while I held on to this idea: "The jobs of the future combine knowing something with doing something (i.e. with your hands) in a specific location". I'm starting to get apprehensive about this as AI gets better and better,
Peter, you've fallen for a common fallacy about LLMs and learning. A specific LLM is locked into the statistical weighting generated by that version's training set, once those values are established the only variations in response will be from that interaction's history or random chance. If a company adds a reference database to establish limits for answers things change slightly and answers may change as the references are updated. Otherwise LLMs cannot 'learn', do not 'understand', cannot 'reflect' when they generate a response. The output will be statistically likely and grammatically correct. Accuracy and truth are not concepts in play.
That is why all the big players want to build 'AI Factories' to do the heavy lifting to train bigger and bigger new models so they have product to sell.
I think you know formulaic writing is nothing new. 10+ years ago my daughter loved reading the Rainbow Magic Fairy book sets by Daisy Meadows. Always 7 books, always the same plot (Human girls meet fairy and one is stolen in Book 1, Book 2-6 random adventures, Girl returned and all are saved in Book 7) with the same villain. She lost interest when she realized the books were only separated by a few find/replace cycles..
Those jobs not only can't be replaced by AI in the future, they can't be replaced by a Call Center drone in another country now. That is part of why males are going to trade schools instead of a college/university.
When a plumber makes up a joint and tests it the results don't depend on DEI or the instructor's political leaning.
The money quote-
"Unless you're in the top 0.1%, you're a peasant," Gawdat said. "There is no middle class."
Formulaic - the reason I stopped watching TV shows by the time I graduated from high school.
"Self-driving trains are common in large subway systems,"
As a freight engineer, we refer to subway type 'trains' as toys. The conditions and variance involved with a short, light, over powered, over braked, relatively flat, and consistent subway operation is the most simplified rail operations one can find.
That being said, current computer operations on American freight trains is expanding significantly.
12 years ago we had computers that 'operated' the train and we had to watch and frequently take over control because it wasn't operating correctly. Or was frequently not working at all.
9 years ago we had computers that watched us operating the train and insured we were operating it correctly - putting on the brakes if we did something wrong.
6 years ago the combined the two and now we have one system running the train and the other system watching to make sure it operates correctly. We still have to be around, currently, for the times it's not working correctly or for the circumstances it was designed to work at all.
Now we have the above, plus interior cab cameras that are scanned by AI in headquarters real time to identify any rules violations (out of rules books thousands of pages long) in order to automatically initiate discipline procedure.
I'm just glad I only have a few more years before I retire.
I find myself wondering about any statics from Great Britain because GB is in the middle of an invasion to change the culture... this is not an external invasion either, it's from within.
Are their numbers accurate or is this part of the invasion?
Elon has a new line of autonomous robots to take care of the bathroom-cleaning and wastebasket-emptying tasks.
I'm getting flashbacks to Heinlein's "The Door Into Summer."
AI os coming for Games Workshop and other miniature wargame figure sellers, and that means toy companies too. People can now run promotional images of toys through an AI and get an stl file to 3d print the figure. Sas soon as GW put out images of a new figure recently, redditors were 3d printing it thabks to AI.
Right, our only hope is for the bubble to burst becauae the old retards in congress won't ban it to save society, and if millenials got in they would mandate everyone have an AI chip put in their brain. The only hope is that the electric grid goes down for good!
Not when they make AI able to write code, now coders out of work. And when AI can make a 3d model of any image, sculpters out of work. You people are stupid. "It can only do what its programmed to do" is not gonna save anyone's job.
I just do not see it for software developers of anything complicated. I’ve got 1.4 million lines of Fortran and C++ that I am dragging around the place. I cannot imagine an AI figuring it out.
Vibe coding is the new hot craze, somebody in a business unit tells the AI what they want to do and the 'bot spits out black box code that maybe sorta kinda does what they want, but without any way to audit if it follows business rules like PCI, actually protects the data with encryption and backups, or can pass the Security team's penetration and vulnerability testing.
Sounds great until the app deletes your production database because of reasons.....
Post a Comment