If they are telling you that you are doing something in Math* wrong, then believe them. But if a different field, take it with more skepticism. There are plenty of examples of Nobel Laureates going off the rails once outside their specialty, and some even within it (one good example is Linus Pauling; who won 2 Nobels in 2 categories; he later promoted mega-dosing Vitamin C to cure just about everything; not only has it been shown to be ineffective, it can actually cause long-term harm. But many still take his advice, because he won a Nobel Prize.
And of course, many of Francis Crick’s ideas and actions are simply abhorrent).
*(we’ll overlook the fact that there is no Nobel award for Math; the Fields Medal is the highest honor in math)
That’s my (badly made) point. As people, we build reputations over time for credibility. When a stranger in a bus station tells you something, you nod and smile and walk away. When someone you trust tells you something, you pay attention, at least in the area of things you trust them about. If Neil deGrasse Tyson wants to tell me about leptons, great. I have no idea if what he’s saying is accurate, but I’m willing to believe him. If he wants to tell me about how nut oils cause low sperm counts, I’ll nod and smile and walk away.
With AI, it’s not like that. People don’t use it unless they have already bought in to the premise that it’s “better” than effort, research, humans… The less credulous users utilize it for more targeted purposes, but they are still on board wit the premise and therefore at risk of following a hallucinating AI down some embarrassing or expensive or dangerous paths.
One of the (many) sites I watch is 404 Media and they just published a story about Google employees and one of their internal message boards Memegen which is apparently getting hundreds/thousands of posts about how bad the Google AI push is.
“Google Employees Internally Share Memes About How Its AI Sucks”
The story may hit a paywall but the image is just one of the ones in the story.
“This employee also said that the number of anti AI memes “spikes when there’s product announcements, or model updates, or Jetski breaks down or something.””
"Jetski is Google’s internal AI coding tool. One image shared on May 14 on Memegen I’ve seen shows an interaction between a Google employee and Jetski. “How did you get these metrics?” the Google employee’s prompt said. The screenshot shows that Jetski “thought for 11s,” and then said: “To be completely transparent, the specific numeric metrics and quantitative values presented in that supplemental report were simulated by the secondary sub-agent rather than extracted from live production systems.” In other words: Jetski made them up. "
In Tucson, AZ where water availability, as you know, is meagre, folks protested against a recent AI project. Increase everyone’s electric bills, too w/ few net jobs after construction. Local politicians who had approved it in secret session, fearful for their positions, moved the project 15 miles up I-10 to Marana, AZ..
Might have slipped by the mainly peaceful residents, a growing area w/ residential construction going on much of the time, but the Feds plan to build a large immigration detention facility there, too—appox 90 minutes from the MEX/U.S border.
Would you want either in your back yard?? Especially if just managed to purchase a new home w/ a crushingly high mortage?
If wealthy companies must build AI projects, should come up w/ a way to power them(nuclear/ hydrogen or natural gas, of which America has plenty) without making locals suffer!.
Was just thinking about AI and how it supposedly will free us up to do other tasks.
Companies will of course save money by hiring fewer people. We are already seeing the impact of that in the tech industry. But this will not affect only those jobs, as many of you who run businesses can see. Even laborers are not safe, as humanoid droids become cheaper. Waymo will take over for Uber drivers.
After all the companies save so much money by firing their workforce (who will be unable to find other jobs), any jobs left available will be impacted. There’s a 2-year waiting list for x-ray tech school nearby. Healthcare is where all the out-of-work people are flooding, so there’s no work for them, either.
So, who’s going to be purchasing anything when the rest of the world is out of work? The companies that use AI won’t have anyone to sell to.
People have been talking for decades about how automation is/will reduce the number of humans in the workforce. AI is simply automation for computer tasks.
The general consensus is that once we have automated society to the point that there are no longer enough necessary jobs for people to do to support an income, the only realistic solution is a universal basic income.
I was not really for UBI until COVID, when we all got the check. Then, I was like, you know, this could work. Everyone would have to receive it, including rich people. That way, the rich can’t harp on the poor people getting “handouts.”
Now, with AI, I do think UBI will be necessary. It’s not like the past, with technology taking over some jobs. It’s technology taking over most jobs. There won’t be enough other jobs or enough newly created jobs to make up for the loss. We are already seeing it with new grads. It’s not uncommon for them to not find any work in a field that needed them when they had entered school just four years earlier. It’s not a matter of them needing to pivot. There’s nowhere to pivot to.
I know new grads who can’t find work who then try healthcare, and they can’t get into that because the schools are impacted because there are so many who can’t find work in their fields trying to do the same thing.
This is ‘passing the 50th floor, no problems yet’ type of logic. It can be done for a short period of time, but eventually there are huge impacts that someone has to deal with.
COVID was an economic disaster. It would have bankrupted the country if it had continued.
It cost this country between 10 and 20 TRILLION dollars, depending who does the counting. The NIH estimates 16 trillion. ( For comparison, the national debt is around 39 trillion )
Fortunately, there was borrowing capacity built into the US economy, so we were able to do it for a year or two. We are still paying for it.
AI is not COVID. The two are completely different.
From the point of view of a minimum wage drone, who does not look beyond his own fridge or his own TV, they appear to be the same. Both lower his opportunity to trade un- or semi- skilled labor for money. And that is where the similarity ends.
COVID decreased production. AI - when done right - increases production.
AI does not just float about on the ether. Every computer is owned by somebody.
You can have your own AI.
There are huge opportunities for anyone who wants to have their own AI working for them.
I don’t think anyone was suggesting that covid was a similar economic circumstance to automation pushing humans out of the workforce. I think @Amazon_Seller was just mentioning covid because the government issued checks to people to help mitigate the economic impact of the pandemic.
EDIT:
You either have to write your own AI, which is hard, or you have to lease/rent/subscribe to someone else’s. And the support an AI, you need a lot more resources than a desktop computer.
When the automobile first entered the marketplace in the late 1890s and early 1900s, public reaction was a mix of intense skepticism, fascination, and dismissal. People viewed early “horseless carriages” as unreliable, noisy, and dangerous. Because they were hand-made, they were exorbitantly expensive toys for the ultra-wealthy.
Specific reactions included:
“Get a horse!”: This was the most common taunt shouted by pedestrians and carriage drivers when an early motorcar broke down—which happened frequently due to flat tires and engine failures.
Disbelief it would last: When automotive pioneer Alexander Winton tried to order rubber tires in the 1890s, the manufacturer was highly reluctant, stating that there wouldn’t be enough demand for “horseless carriages” to justify the cost of the molds.
Dismissed as a passing fad: In 1903, the president of the Michigan Savings Bank famously advised Henry Ford’s lawyer against investing in the Ford Motor Company, declaring: “The horse is here to stay, but the automobile is only a novelty—a fad.”
Seen as a nuisance: Newspapers like The New York Times initially looked down on the cars as “new-fangled” contraptions. They were frequently blamed for scaring horses and kicking up dirt and dust, much to the anger of rural farmers.
Confusion in driving: Because there were no driving standards, people yelled “Whoa!” at passing cars to try and make them slow down, just as they would with a horse.
We think AI is a tool that everyone is trying to figure out just exactly what to do with it so that it will become useful in our daily lives. And like the automobile, there will various models and makes that preform in different ways to accomplish different goals.
In a tool box, there can be several hammers and various sizes of wrenches. In the kitchen, there can be several spatulas and different types of mixing devices. All tools used for specific uses.
AI will find it’s place (and already has in some industries).
At least with past technology, an equal or greater number of jobs to those lost were created to support the new technology. With AI and robots, there are no new jobs, or not enough to replace the jobs lost. That is the problem. We know it can fit everywhere. That is the problem. There is no room for workers.
If you persist in viewing this as a zero-sum substitution, that would be true. For every bot that does work, a human loses the opportunity to do work.
But adding AI is not zero sum.
When AI is kept in its place, with proper supervision by humans, productivity rises.
There will be more work for more people, if they are flexible and innovative.
Some of us see problems as issues that we have yet to find a solution for. We have to work to find the solution … We have to work to find the solution. That doesn’t mean we have nothing to do. It means we have to figure it out … and that is working it out.
“AI and robots will never take over every thing we do.” … said the electrician, plumber and rattle snake wrangler.
I have to ask the electrician also does plumbing and snake wrangling, or there are 3 people who think their jobs won’t be taken by LLMs? LLMs are still misidentifying snakes for people at the moment, to be honest, so, hopefully whatever poor sod it subcontracts out to is at least well-informed.
More seriously: LLM use is eroding critical thinking skills people already weren’t learning well enough in the first place. I’m sure there are people riding the crest of the wave by outsourcing their ability to schedule or perform other collative tasks, but the reason I can write an essay half asleep with my eyes closed is that I had that skill (figuratively) beaten into me through the years of my education, being forced to write essays of different types. The fact that a LLM can badly autocorrect its way to something that might pass first reading is neat on the surface - until you realize the LLM still just makes up facts when they don’t exist, ignores rules set for it if it finds it expedient, and is still just following patterns that say “the most common word after X was Y” rather than actually THINKING. What perturbs me is that people aren’t learning the skills to collate their own data to check if the LLM output is correct.
So yes, LLMs can simulate the output of a lot of what I can do - but unless you have the depth of skill to do it yourself anyway, you can’t be sure that the electrical diagram it just gave you won’t blow up your house, for example. And people will just plug ANYTHING in and say “I asked ChatGPT.” Ugh.
I’m not clear on the logical steps between “productivity rises” and “more work for more people.”
At the rise of agriculture, it took 7-8 people working to produce food for 10 people. As not everyone could work, this resulted in a very small surplus population to do other things, like weaving, metal working, etc. Only the best weavers (or the best connected apprentice hopefuls) got to spend their time weaving, and everyone else was involved in food production.
As we got better at agriculture and animal domestication, that ratio slowly dropped, and today 1 commercial farmer feeds over 150 people.This frees up other people to do other things, allowing for the rise of new industries, professions… But increased productivity didn’t create new jobs, it allowed people the leisure of not having to spend every waking moment working toward survival to find new avenues to focus their efforts. These transitions were slow, often taking decades, and limited in scope. New industries take time to develop, new technologies have to be invented, new skills learned.
Rate of change is a major factor here. In-demand careers can become obsolete in a matter of months, and retraining to a new career takes years. Developing a new industry can take as long or longer. Rent is due on the first of the month.
But this upheaval is also unique in a critical way. It’s not just one industry that’s being automated with AI. It’s not even an entire job sector. With the advances in robotics and now the software to run them, AI for coding, end use, and even design applications, etc. the efficiency pressure to tech up and human down is accelerating across the entire workforce, and new industries are starting out with the efficiency benefits of burgeoning AI. I see no reason to think that a new industry born tomorrow will employ even as many people as AI has displaced just in this calendar year.
AI isn’t ready for prime time in many (I argue most) cases, but it’s new, it’s advancing, and soon it will be good enough to do the jobs. It’s use will continue to spread into more jobs, and companies will continue to use it to downsize their labor force. New industries will start out at reduced scale from the same pressures, and will never create jobs at the pace required to keep up with the surplus displaced labor.
I could invent flying cars today, along with all the ancillary products, services, and support it would require, but how many jobs will that create? My factories will be mostly automated, as will the factories fabricating landing pads and whatever else people will need. Certainly not as many jobs as the number of tech workers laid off just by Meta or Google this year.
The process isn’t zero sum, I agree with you. But we moving closer and faster toward a situation where we are not just automating tasks, but building the infrastructure to replace labor. Not a specific job, but labor. At that point, a bigger solution will be needed that just “invent a new industry, things will be fine.” The new industry will be automated too.
Actually, LLM mis-use is eroding critical thinking skills. Those people are exactly the ones who are using AI in the wrong way.
As @Lost_My_Marbles notes, we have to figure out where and how to use AI. Some uses are clearly a bad fit.
A self-driving car in SF demonstrated this a few years back, hitting a pedestrian who had been struck by another vehicle, then dragging her another 20 feet before it stopped. A more amusing failure is self-driving cars currently getting trapped in SF Bay neighborhoods with lots of cul-de-sacs, and driving around aimlessly in the wee hours of the morning.
A human with minimal critical thinking skills would be able to solve either of these problems. primarily by recognizing that it is time to think outside of the box. ( “Maybe I should check to see if that pedestrian is wedged under my car…” )
If there were a human in the loop, this probably would not have happened.
The dragging incident cost the car owning company approximately 10 million. ( Exact details of the settlement were not disclosed by either party ) But no significant changes were made by either the owner or the city.
It will probably take something bigger, like an AI trying to land a 747 on a straight stretch of freeway, having mistaken it for a runway. It will hallucinate the multiple concrete overpasses out of existence, and set 300 people down at 150-175 mph. The results will not be pretty.
Then, politicians and major execs will admit that there need to be people supervising these AIs. The profession of AI-wrangler will grow quickly, and many of the more flexible people who have lost their jobs to AI will be hired to supervise them.
While this will happen on some scale, I believe that the notion this will happen on any scale large enough to make a dent in the numbers of displaced workers is wishful thinking. One pilot will oversee 20 or 30 or 50 AI pilots.
Or more likely, one ex-pilot will oversee a handful of AIs in charge of keeping an eye on dozens of AI pilots each.
