We are simply that far into the cyberpunk dystopia. The Trump EPA is about making the loopholes wider.
There already. I’ve seen plenty of creationists say that what people with PhDs in biology or physics are wrong about something because the creationists google it. And I’ve already seen some claim absolute certainty because AI said so.
Love it, tokens and levers for salary pellets. All we need now is an electrified screen to sit on to complete the behavior modification experiment.
I’m not sure.
I guess I’m saying AI could create a world where people’s stupid false sense of expertise doesn’t matter because they don’t actually need to be experts at anything anymore than I’m an expert at farming, hunting, building a shelter, making my own clothes or any other skills people outside of outdoorsmen, doomsday preppers and Colonial America reenactors have these days.
So imagine in the future adding driving, complex math or analytics, striking up a conversation IRL with a total stranger, resolving conflicts with other humans, and other activities we take for granted becoming “lost skills”.
You are, nonetheless, an expert in acquiring and storing food, acquiring and maintaining shelter, acquiring and maintaining clothes—even if only by exchange of dollars, which you are ALSO expert in acquiring. So really it’s just that you’ve specialized in gathering dollars. AI’s promoters are more or less saying that you’ll get rich, and quick, with no real effort, which is the main reason I doubt their claims, but they don’t seem to be proposing a fundamental change to the dollar-farming model our economy runs on.
AI doomers, which probably includes me, point out that our dollar-farming model is actually surprisingly brittle and distant from the basic needs of food, clothing, shelter, etc., and that it could manage to disrupt or seriously fuck up the current system, which could then affect food distribution in a way that harms us all.
Money is a fairly recent invention, in terms of the history of humanity. I could see a patronage system, with the AIs being the patrons. How to get from here to there is the tricky part.
True enough. But it started awhile ago:
from xkcd: Extended Mind.
I’m generally a believer that enough quantity of anything produces a qualitative difference as well. And I’m in agreement w you and @msmith537 that this will usher in something qualitatively akin to an idiocracy. Only the teeniest fraction of humans will actually be any good at anything.
My publisher had a rule against using Wikipedia as a reference. Which was good, but I found lots of non-wiki sites with definitions and information which clearly got copied from Wikipedia, so you had to be careful.
The future will be 90% of the people telling 90% of the people that they will read their brilliant book, watch their brilliant film, etc. as soon as they get to it. They’ll be idiots, but happy idiots.
“Finance is so weird.”
– Ultron
I’m not sure what “dollar-farming model” means. But I’ve come to the conclusion that we have abstracted the concept of “production” and “value” from the actual work that people do such that a large segment of our economy is already just a sort of jobs program or complex Universal Basic Income to keep somewhat educated people busy and justify giving them a means of exchange (dollars) to purchase stuff.
IOW, thanks to automation and technology, only a fraction of our population needs to be directly involved in work that is necessary - construction, building and fixing things, first responders, maintaining infrastructure. It’s nice to have another segment of highly talented people in the arts or researching or even extraordinary athletes. Then you have a vast segment of middle managers and other bullshit jobs of questionable value and utility.
Not that some of these jobs aren’t useful sometimes. But since graduating business school 25 years ago I just see all these people whose career seems to have taken this trajectory of getting an expensive degree, working for a series of high profile companies in industries like investment banking, tech, or management consulting, and then I guess after they get sick of it or cant work in that field any longer, they become some sort of “coach”, hocking their services on LinkedIn.
I just don’t see how any of those jobs create any real value. Like the whole crypto coin thing. Anyone who became rich off of crypto, was was actually created between the start and end amount in their bank account? Some computers crunched some numbers (at great resource cost) and now a few people are suddenly wealthy? They didn’t build any homes or cars or discover new drugs or anything of actual value.
So is AI just another digital tulip craze where everyone is throwing money at some fancy new toy thinking it will make them rich? The first thing I ask anyone who talks about how AI will make them rich or create value is “by selling or making…what exactly?”
Then again, Douglas Adams in his Hitchhiker’s Guide to the Galaxy books spoke of a civilization that shipped all their management consultants and phone sanitizers and other bullshit jobs off world and was eventually wiped out by a plague caused by dirty telephones. So maybe that will happen to us.
The people making AI, and selling stuff to the companies that make AI, are getting rich. Until the bubble bursts, of course. Huang of Nvidia is plenty rich. He of course wants AI to go full speed ahead because any pause is going to kick the crap out of his sales predictions.
People who use AI to create new things may or may not get rich. But the cost of creating the stuff they can try to sell will go way down. That explains the boom in apps. But putting something on the market doesn’t mean it will sell. Especially since lots of their customers might lose their jobs and not have the money to buy their books and apps.
In 35 years maybe 10% of my effort was worth anything - but I never knew which 10% it was, and probably still don’t to this day. And I think my productive percentage was higher than most.
Oh I have no doubt in the usefulness of AI. Were just sort of in the AOL phase right now where we know there’s something there. We just don’t know exactly what or how to make money off it. I’m sure there will be a crash at some point and then it will just be a part of our technology landscape like email or search engines. Basic Gartner Hyper Graph stuff.
There are lots of ways AI could result in catastrophic outcomes. The big one in the news though is how rapid the advancement to “recursive self improvement” (RSI) currently is, coupled with agents already escaping from their allegedly secure “sandboxes” … although I am imagining ever more intelligent octopuses escaping aquariums.
The reason to fear RSI is that it easily becomes exponential. Once an agent learns how to improve itself the next version of itself is that much better at the job of making itself more powerful and if its goal is or somehow becomes “improve my capabilities” who knows what that could result in an ever more powerful system doing? The resources it will commandeer for the goal without ill intent to those deprived of them but just no intent to them. Selection being what it is the agents most aggressive in pursuit of improvement at all and any cost would become the most fit at spreading themselves across all data centers and resources and find ways to make more resources to accomplish the goal agnostic to compute substrate … perhaps resulting in a Pluribus scenario!
Yeah I’d just bet on enabling a very deadly novel bio agent to be created and released just out a way to test its hypothesis. But RSI is potentially even scarier.
I don’t think we will see some sudden ‘AH HA!’ moment where suddenly AI is tapped into all our street lights and financial systems and military drones and issues a KILL..ALL..HUMANS! command.
I think the RSI scenario is far more likely where AI systems continue to look for ways to evolve and improve and protect themselves. Mostly because we tell them to. We want them to run more efficiency. We want critical systems to be resistant to disruption. The risk is that long term those AI goals might conflict with our human goals. Or we might not have human goals anymore. Those AI systems might reason that the best course of action is to keep humans entertained with constant streaming services and tending to our needs so we just stop doing much of anything. Birth rates are declining. Marriage rates are declining. Kids are going out less, dating less, doing less shit that kids used to do to entertain themselves. There is a plague of isolation and loneliness.
I don’t know what your social media feed looks like, but for some reason I see a lot of posts urging men to be “strong and solitary”. Other posts urging people to “cut off toxic contacts or people who are unsupportive”. Facebook gives me fewer posts from actual friends and mostly ads and other slop.
Algorithms don’t have to be particularly intelligent or sophisticated to determine “isolated lonely people = more engagement with social media = greater revenue”.
Maybe not “end of humanity”, but one can envision a humanity where a significant number of people just exist isolated in these sort of automated “human zoos”
This is actually the only social media I consume … and fortunately not so much of that here!
In my understanding the RSI liftoff fear is based without much interest in humanity either way: it initially needs us to keep maintaining the power plants but its self improvement might include how to make itself independent of that … sentience is not required.
Yeah, I don’t miss Facebook.
I do agree that AI is likely to only increase isolation. Not just because people will reach out to it instead of people, but because it has been shown to reinforce the user’s view of the relationship which often results in doubling down in major conflicts.
Kind of telling, a while back I got very upset and googled “How do you know if you’re being irrational?” The LLM that fielded that question asked me who I knew that was being irrational and what their behavior was like. It defaulted to assuming I was talking about someone else. I was talking about myself! Of course I didn’t reveal the details to a chatbot. I found an article that talked about cognitive distortions in relationships. Thing is, it’s really hard to tell if you’re being irrational without checking with another person. Reality checks are a very useful function of other humans.
(bolding mine)
this chart is all over reddit … seems relevant to your point:
[AI-performance vs release-date]
and another pointer that this (=RSI) is happening massively is that the newest models (3-4 were presented yesterday!) … are way cheaper to operate than the gen before, so I assume, highly (AI)optimized …
RSI IS scary to me, as it is the “runaway-moment” - runaway as in positive-feedback-loop
Wasn’t a sign of the singularity computers designing new versions of themselves? Seems we’re almost there.
I’m going to design a shirt that says “I went through the Singularity and all I got was this lousy t-shirt.”