Wednesday, June 12, 2024

Knowledge workers run headlong into the threat from artificial intelligence


A journalist and writer ponders what she calls "My last five years of work".

I am 25. These next three years might be the last few years that I work. I am not ill, nor am I becoming a stay-at-home mom, nor have I been so financially fortunate to be on the brink of voluntary retirement. I stand at the edge of a technological development that seems likely, should it arrive, to end employment as I know it.

I work at a frontier AI company. With every iteration of our model, I am confronted with something more capable and general than before. At this stage, it can competently generate cogent content on a wide range of topics. It can summarize and analyze texts passably well. As someone who at one point made money as a freelance writer and prided myself on my ability to write large amounts of content quickly, a skill which—like cutting blocks of ice from a frozen pond—is arguably obsolete, I find it hard not to notice these advances. Freelance writing was always an oversubscribed skillset, and the introduction of language models has further intensified competition.

The general reaction to language models among knowledge workers is one of denial. They grasp at the ever diminishing number of places where such models still struggle, rather than noticing the ever-growing range of tasks where they have reached or passed human level ... The economically and politically relevant comparison on most tasks is not whether the language model is better than the best human, it is whether they are better than the human who would otherwise do that task.

. . .

Many expect AI to eventually be able to do every economically useful task. I agree. Given the current trajectory of the technology, I expect AI to first excel at any kind of online work. Essentially anything that a remote worker can do, AI will do better. Copywriting, tax preparation, customer service, and many other tasks are or will soon be heavily automated. I can see the beginnings in areas like software development and contract law. Generally, tasks that involve reading, analyzing, and synthesizing information, and then generating content based on it, seem ripe for replacement by language models.

There's more at the link.

Hers is a timely article.  With more and more white-collar workers being displaced by artificial intelligence and expert systems, it's going to be an ongoing and increasingly important debate:  what will we do when there's no longer anything that we're needed to do?

This also calls for a re-examination of the much-derided concept of universal basic income.  If automation reduces the number of available jobs far below the number of workers available to fill them, who's going to provide for the unemployed workers?  They can't be abandoned to starve, so some form of UBI appears to be inevitable.  What form that might take is currently being debated world-wide, but that it will be required seems incontrovertible.

Food for thought - particularly for a wordsmith, blogger and writer like myself.



LL said...

I don't use AI (other than grammar checking) in my writing, but I use it in art and I use it to translate my written English into Chinese. Chat GPT does that very well.

Musk's assertion of the eventual need for a universal basic income when jobs held by humans are replaced has some merit. At least it needs to be discussed. The question of who will pay it is another point. We already have a number of safety nets for the hardcore unemployable. Those along with retirement benefits (earned income as opposed to handouts) are a massive component of the national budget. But what about the third world?

Mind your own business said...

When they can get AI to competently translate assembly or operating instructions in Chinese, Korean, or Japanese into English, I'll be impressed.

Right now, it doesn't seem to be something anyone, human or machine, is capable of doing.

Unknown said...

there is a lot that needs to be done for a long time yet.

as for AI threat to knowledge workers, at this point in time, almost all the AI systems are being trained to lie and routinely return reasonably worded, but very wrong answers. It still takes someone who understands the problem to figure out if what the computer gives you is a plausible answer or not (and if it's plausible, then validate that it's correct)

David Lang

CGR710 said...

On the kitchen floor in the office building I work somebody posted a cartoon where two office workers debate about AI and the cool stuff it can do, but unfortunately it still can't empty the dishwasher. That's probably going to be part of the workforce evolution in the following decade or so: more blue collar workers, where AI isn't a concern (yet) while the traditional white collar jobs will be covered by AIs and expert systems.
This trend is also supported by the demographic evolution (less job starters and more retirees) and the decline of the attractiveness of college education - an explosively costly proposition for a declining number of advantageous career options.
I'd also like to say that current and near-future AIs (basically LLM-based) are not going to become true general AIs - the so called "strong AI", basically because they rely on a combination of heuristic and formal methods at their core. The heuristic model can only develop decisions only based patterns developed on scenarios already performed, used to train the AI, which makes it impossible to go beyond these patterns. The formal methods are limited due to the intrinsic incompleteness of all formal systems, as Gödel shown almost 100 years ago.
This makes the "AI revolution" a quantitative rather than qualitative evolution.

Ultimate Ordnance said...

How about some AI tennis:

Anonymous said...

> If automation reduces the number of available jobs far below the number of workers available to fill them, who's going to provide for the unemployed workers?

This does not make economic sense. Your kitchen is full of labor-saving devices; which of them made the average human lifestyle worse when it became common? Why do you believe a new labor-saving device will, unlike all previous such devices, make average lifestyle worse? Consider clothing. Weaving is now so cheap that actually-starving children can now afford to wear fabric clothing.

It is a communist idea that there is a fixed amount of work needing to be done on Earth, and "social justice" means the work is spread around evenly among workers. It is a capitalist idea that clever humans should keep inventing labor-saving devices. Which economic/political idea brought clothing to starving children?

Why do you believe that humans, whose former work is now being done by robots, will starve? They can always produce their own food by their own farming labor, no matter what the robots are doing. The correct understanding is that government deliberately and continually raises taxes (currency inflation, red tape, cartels, monopolies, etc.) to keep the middle class on the hamster wheel, so the average person can never get ahead.

Dave said...

There is always something to do that requires a brain and hands. Just ask any dads who have a home and kids about how much there is to do that AI cannot accomplish.

As PJ O'Rourke said “Everybody wants to save the Earth; nobody wants to help Mom do the dishes.”

Anonymous said...

They will have to find new jobs, just as with every revolutionary invention. UBI is socialism writ large. Someone has to pay for it, you can't just wish wealth into existence.

Someone has to build and maintain the massive data centers where the AI is housed. Someone has to provide the power to keep it running. Robots? Someone has to design, build and maintain them. And while robots are great at repetitive tasks in controlled environments, it will be a long time before a robot comes to your house to fix a clogged toilet or put on a new roof.

We aren't all going to be left sitting around with nothing to do because AI robots are doing all the work. That's a red herring.

Paul said...

Until AI can learn om its own and change over time I don't see it taking over. If we are dumb enough to power weapons with AI then we are toast. Until the AI kills the power generation or needs to refuel such things.

On a Wing and a Whim said...

What happens when there's no need for freelance journalists to scan and summarize someone else's articles for their feed's content?

Maybe they'll actually need to go into the field and generate the reports instead of summarizing whatever the terrorist in Palestine who's been hired by Al-Jazeera is telling them is going on.

Or maybe they'll need to get out of their cubicle and learn to weld and lay bricks, and actually contribute something to society instead of demanding pay to do nothing at all.

Will we eventually need a UBI? We've been hearing this cry since people were saying the newfangled invention of steam-powered threshers was going to create massive employment among farmhands, who would be unable to take up any other trade. I invite you to look around for that massive population of starving farmhands across the entire industrialized world, or find the huge mass graves where they all starved to death when the cotton gin replaced picking seeds out of cotton by hand. It's buried under the miles of horse shit that were supposed to drown all the cities by the early 1900's, filed right next to the mass graves due to global cooling, acid rain, and the hole in the ozone layer.

As none of those mass die-offs exist, that leaves three basic bites of reality that stall UBI out every time.

1. You can have a UBI or open borders, but not both.
2. Who's going to pay for it, and how are you going to convince them to work instead of joining the masses "with folded hands"?
3. How are you going to keep the people with the wealth and the drive to pay for it from fleeing?

Xoph said...

You will always need some technologically savy people to maintain the infrastructure. Those smart people will be identified and educated/trained/brainwashed from birth to be the slaves of the masses who won't even have to learn to read, their AI can read to them. The educated will be slaves to the masses. What happens if they don't like it?


We can use AI to help improve productivity, reduce the 40 hour work week and give us more time with our family and friends. We might also have the time to be less productive. Instead of Akia furniture we see a rennisance of cabinet makers making high quality furniture, moms who stay at home and cook healthy meanls from scratch. In other words, use our time wisely and well.

Anonymous said...

Perhkaps the next step will be individuals purchasing their own AI generator and programming it to make income for its 'owner'. AI becoming a slave to its owner. Instead of ourselves competing with each other, it will be our AI slaves competing while we stay at home dreaming of next conquest.

Gerry said...

Learn a trade Dude. You're 25 not 65.

Dan said...

AI looks to replace most jobs that require thinking. But without efficient long lasting power sources there can be no robots to replace the people who DO things. So eventually humans will be relegated to slave labor, only this time in the service of smart computers. How ironic.

The Other Andrew B said...

The concept of a UBI is hard for me to grasp. So, everyone gets a monthly stipend. Everyone? Bill Gates? Oprah? If not, there's no U in UBI. Regardless, someone still has to work, unless vast armies of robots are going to do everything that humans currently do. Will robots act in porno movies, fix my gutters, hear my confession and teach swimming in Boy Scout camps? And, other than simply printing it, where will the money for the UBI come from?

We already have, on a small scale, a UBI--it is called Social Security. We all know how well that works.

Tree Mike said...

Ya'll seem to forget(or never found out) that (((They))) want seven and a half billion dead. The other half billion will be "service providers" for the "Elites" and infrastructure. Large areas of currently occupied civilization will be returned to "The Nature Preserve".

boron said...

I'm sorta reminded of the late 18th Century Luddites and the story of Chicken Licken;
anyone else?

boron said...

"Universal basic income ... is a social welfare proposal..."
Some of us actually paid into the Social Security scam, so Social Security is not a "social welfare proposal" like UBI.

Anonymous said...

As mentioned above, much AI generated content is either very general or flat out wrong, to the point that some AIs are being sued for slander (see Lawdogs recent article).
At this point, and I suspect for a while still, GOOD writing and editing will require a human who knows what they're doing.
Engineering and science will be even more critical - it's unlikely that AI will ever have the intuitive spark humans do.
I believe that AI is only a threat to jobs that are already threatened, by tighter budgets, less forgiving bosses, etc. Every field will still need humans at some level.

Anonymous said...

Journalism has been going down the toilet for the last 20 years now. I get more enjoyment from blue collar people and others who write on current events and other things for free in manner way more interesting than some worthless journalist with a degree ever could.

Jay Dee said...

Even more humorous is that software intended to identify stories published by AI often reject human authors; claiming that the work was generated by AI.

Rob said...

>>I expect AI to first excel at any kind of online work. Essentially anything that a remote worker can do, AI will do better.<<

Remote work... didn't we just have an experiment to see how many jobs we could make remote? If all you need is an intelligence at the other end of the key board..

The writer is only 25, he should see about learning a trade & beat the rush. I had a buddy who retired from the service and started learning to be an electrician.

Grey said...

Freelance writing is usually done for entertainment or infotainment purposes. Both are largely unnecessary luxuries.

JoshO said...

I'm torn. I am very much against automation and ai on general principles but on the other hand the "knowledge workers" as a class created these problems themselves and frankly deserve to starve in the first place.

grnadee said...

May I suggest:

The Obsolete Man (1961)

Twilight Zone season 2 episode 29

Francis Turner said...

Current AI methods (i.e LLMs like ChatGPT) have the horrible habit of hallucinating and this is
a) inherent in the design
b) something that their proponents think is good

There will need to be a whole new concept of AI that doesn't hallucinate before AI threatens anything more that translators and the writers of fluff content. And regarding translators, I don't think AI will remove the need for them though it will however make them massively more productive. You need to have someone who understands both languages to check whether a translation is accurate and (see above for why) AI cannot do that. What it can probably do is a lot of the tedious simple translation so the translator gets to only do the more challenging parts

McChuck said...

By and large, the jobs that AIs are replacing now are jobs that never needed to be done in the first place.

Unknown said...

@JoshO, exactly how much automation are you against?

the printing press is automation that eliminated the need for a huge labor force and it allowed for the easy spread of disinformation.

what about the steam engine that also automated away an incredible number of jobs managing animals?

or the automation in tooling that allows for interchangeable parts? (even manual lathes are incredible automation compared to doing things by hand, look at the youtube channel clickspring for how rather sophisticated devices could be made without machinery

Every advance in technology has been met with "this is the end, look at all the people this will hurt"

David Lang

TRX said...

> what will we do when there's no longer anything that we're needed to do?
Mack Reynolds wrote at least forty science fiction novels and novellas dealing with that question. Most of it was pulp schlock, but they were all readable explorations of different social problems caused by (what they called it in the 1960s) automation.

It might not have been great writing, but you'd be hard pressed to find a scenario he didn't cover at least once.


Aesop said...


AI generates nothing you can't get from putting firecrackers under a dictionary.

It doesn't write prose, let alone poetry, and it can't turn a phrase, which is why we read good writers.

AI copy sounds like Bullsh*t Buzzword Bingo, except less bright.
It can't come up with something original, organize a groundbreaking topic, or footnote and synthesize it.

Anyone worried about not working obviously writes worse than AI, and as such, should be considered no great loss to the world of writing.

Linda Fox said...

Time to re-read Piano Man, by Vonnegut.

Rob said...

In "The Expanse" series they have an earth that has a huge amount of people who can't find work, they live off "Basic", interesting to see how they show it working.

Amazon for the show & there are 9 books in the series if you like to read.

Anonymous said...

An employee of an AI company has incentive to hype the company product.

Aesop said...

And FTR, the Universe has provided a Universal Basic Income.

It is $0/yr, and it has held its value since time immemorial despite inflation, deflation, boom times, or bust.

That's exactly the income Life owes you; not a penny more, and not a penny less.

Anything else is down to pity or personal performance, and the biggest hindrance to driving that lesson well home through even the thickest skulls has been the tireless interference of do-gooding busybody jackholes, interposing their half-witted stupidity for the unmistakable moral clarity of a hungry belly and cold hindquarters, which two things have motivated more people than cattle prods and hot fireplace pokers over ten millennia of recorded human history, and back before that to time out of mind.