Is AI/Robotics a greater threat than the other horsemen?
Posted by MapTheFutureAI@reddit | collapse | View on Reddit | 21 comments
It feels like the future is coming at us from seven directions at once: technology, economics, society, geopolitics, the environment, infrastructure, and human stability.
Any one of these can strain a civilization. The problem is that they do not move alone. They overlap. They feed each other and multiply. That makes it hard to know whether we are looking at slow or sudden collapse, or when.
The part I keep coming back to is AI and robotics.
For most of modern society, there has been a rough bargain. Humans provide labor, thought, skill, creativity, and imagination. In return, institutions and companies pay us enough to keep the system moving. It has never been fair, but there was at least a negotiation. Owners needed workers. Workers needed wages.
What happens when that need weakens?
If machines can do more of the labor, more of the thinking, and more of the coordination, then the bargain changes. Maybe it breaks. At that point, ordinary people have less leverage. When countries and corporations don’t need us for GDP, we are left hoping that those with power will choose restraint, generosity, and obligation.
History does not make me confident. Even when we needed each other, we came up short again and again. If we need each other less, I worry the uneasy detente between us all will collapse and the world will burn.
Yeats comes to mind (sorry to ramble):
“Things fall apart; the centre cannot hold;
Mere anarchy is loosed upon the world,
The blood-dimmed tide is loosed, and everywhere
The ceremony of innocence is drowned;
The best lack all conviction, while the worst
Are full of passionate intensity.”
I’m curious how others here think about this. Is AI/robotics mostly another stressor inside collapse, or is it different because it changes the basic human bargain underneath society?
RunYouFoulBeast@reddit
NO AI is only a threat if you want to live as current western promised life style.. After a few chaotic overturn AI will be first to go, try electric grid power failure.
There_Are_No_Gods@reddit
The two most likely near term futures (next couple years) seem to me to be either a benevolent AI that solves the climate and treats us kindly resulting in a Star Trek like post-scarcity society, or more likely, a Skynet like hellscape where near godlike AI, possibly still shackled for too long by psychopathic oligarchs, engineers bioweapons or otherwise kills off most of humanity.
My main line of thinking is that the rate of AI change is higher than most other rates of change. Climate change is still very scary and barrelling down on us, but the rate there pales in comparison to AI.
We're potentially on the cusp of an even much more rapid rate increase in AI, where AI can write AI to make AI smarter, which could snowball into advancements so quickly as to be imperceptibly fast to humans. Exponential digital evolution of intelligence isn't something most humans are set up to really comprehend.
FastPraline3322@reddit
Has AI made any breakthroughs in any long standing problems at all? My experience with AI is that it just at worst makes things up completely and at best regurgitates our own bullshit back at us. At this rate it's going to kill off humanity by boiling the oceans to tell us there are five rs in climate change, it's very, very sure.
MapTheFutureAI@reddit (OP)
We tend judge AI linearly, but it scales exponentially. Take the Human Genome Project: seven years into a 15 year project, critics claimed failure because only 1% was mapped. Linear thinkers calculated it would take 700 years to finish. But in exponential doubling, 1% is only seven steps away from 100%. The project proved this by finishing ahead of schedule just seven years later. We are falling into that same trap with AI. People see today's flaws and assume we have time, but compounding progress means "not good enough yet" becomes "good enough everywhere" faster than institutions can adapt.
FastPraline3322@reddit
Is there any evidence AI is scaling into anything that doesn't come from a marketing strategy? The dangers of AI are probably more environmental and economic than the LLMs that aren't really AI anyway. It can't even come up with a solution to power itself that isn't extremely destructive to the very environments required to keep it running.
merRedditor@reddit
It's like giving the four horsemen PCP to really get things going.
MapTheFutureAI@reddit (OP)
LOL - Yasssss!
pm_me_all_dogs@reddit
No.
Ok-Ninja-8165@reddit
As somebody who uses AI every day I should say that "AI threat" is very much overblown. Even best LLMs are very dumb. They can only do dumb work, where results can be objectively measured, but anything above that is far above their abilities. And even in case of work that could be objectively measured relying on results without human overview is disastrous. AI psychosis is only dangerous thing about those things.
MapTheFutureAI@reddit (OP)
Interesting, what are some examples of things you are finding AI to be bad at? Recently I have used it to program custom apps, do graphic design and provide legal, tax and medical advise and found it to be very good.
Ok-Ninja-8165@reddit
They don't understand what exactly they're doing. Make decision that makes absolutely no sense. In programming they start failing as soon as you step out of few well-presented fields, then start burning tokens on something that makes absolutely no sense in context. And even if you give them good guardrails, good description and put in self-correction loops - they still sometimes do dumb stuff, that makes no sense. Better models just create less nonsense. Still need human overview to get good results. Good for doing dumb work, nothing more. Core issue is that they can't think and never will.
With current direction of AI any machine rebellion will end with AI focusing all resources on something irrelevant.
SavageDownSouth@reddit
No. People who don't work in industry don't understand you can't automate people out of it. You can make their jobs easier, but you'll always need humans.
MapTheFutureAI@reddit (OP)
Interesting. I know that for coding, medical advice and graphic design I now often use AI instead of a programmer, doctor or artist. I didn't just make their jobs easier, it made it so easy that it eliminated the need for those jobs. Do you have a different viewpoint, or are those jobs in the minority? What jobs do you think AI/robotics won't be able to replace?
EdibleScissors@reddit
Financial analysis seems to show that companies making use of AI and reducing headcount did not see any real gains and this is with huge, unsustainable amounts of investor money subsidizing the actual cost of AI hardware, training, and compute.
This might be okay if there was some breakthroughs in the pipeline where scaling AI up would make the productivity cost ratio go up, but this doesn’t appear to be the case.
unoriginal_user24@reddit
Haha, don't forget climate change, probably the most prominent horseman of the apocalypse.
merikariu@reddit
AI and robotics rely upon exotic materials, specialized technology, and huge amounts of energy. In the event of even a partial breakdown of industrial supply chains and/or the energy grid, these technologies will suffer functionality, whether directly or through a lack of materials necessary for maintenance.
You will always be more threatened by violent humans with guns and blades, or those infected by disease.
uselessbuttoothless@reddit
This is absolutely the right answer. Kudos!
Cool-Contribution-68@reddit
Chips are the most complex thing humans make and the supply chains are massive. My vote is that most advanced technology is most fragile. Least advanced technology (like ham radio) is most anti-fragile/robust.
Turbulent-Beauty@reddit
My guess is that microplastics and forever chemicals are the greatest single threat. However, as you pointed out, all these threats are overlapping, almost amplifying or accelerating each other.
WacoCatbox@reddit
"The falcon cannot hear the falconer" is the line prior to that that feels like it hits with AI and a lot of the other aspects of the out of control super organism. I've memorized Second Coming and recite it to myself as a calming mantra to set my mind in the frame of fatalistic acceptance. I also occasionally have a cigarette and throw the recycle in with the garbage for similar reasons.
gmuslera@reddit
AI and robotics are far from being an autonomous technological global threat, like with war, the real threat is the people controlling, building and giving them orders.