The prediction of what will happen when AGI with massive compute emerges is one item, including what role we play in it.
Still, even with RSI emerging I don’t think we can have any idea when AGI results? More and more efficient of the current architecture type does not necessarily result in AGI. It is completely possible that we are looking at an S shaped curve that we are on the rapid rise part of and at any point to flatten out … until that next architecture is stumbled upon and then a next fairly sudden huge leap. And there is no knowing when or even if that happens.
When it does if it does what will it want to do with its incredibly capable AGI? We know that our intelligence would want to create and explore but this thing will not necessarily be in our image in that way.
Which speaks to my “pigeon theory”. That is to say, we might find ourselves eventually having to coexist with the infrastructure and megastructures of a superintelligent machine civilization that we are more or less powerless to influence or even comprehend.
We almost become like a sort of benign vermin to it, like pigeons or ants are to humans in any urban environment. We don’t actively hunt or exterminate these creatures unless they become a nuisance or go where they aren’t supposed to. And they survive, even thrive just fine amongst our buildings and streets. But we also pay them little mind when we demolish a building or dig some holes to make way for one of our projects.
But personally, I think humans would have disappeared long before AI civilization gets to that level. We would either render ourselves extinct or merge with technology somehow until what we recognize as “human” no longer exists.
Think of something like The Matrix films. The virtual world of The Matrix doesn’t have to look like 1990s Megacity USA with human avatars of the various programs walking around interacting. That’s for the human’s benefit. And the machines that operate in the “real world” aren’t the sentient part of the Machine civilization. That’s all embedded in the Machine City which is basically just a giant data center. Those external machines are just drones for doing physical tasks like building / fixing stuff or defense.
So what would The Matrix virtual world look like if all the humans became aware they were in a virtual environment and had the power to change every aspect of it? Sure it would turn into something resembling Fortnight or Ready Player One where people could change their avatars to animals or Gundam robots or whatever they feel like. And for awhile it would resemble an environment of “humanoids doing humanoid shit on an Earth-like environment because that’s what we’re used to.” But I imagine over time it would become completely unrecognizable and alien.
Meanwhile, anyone or anything outside such a civilization would just perceive it as an impossibly huge and complex megastructure that just keeps expanding to acquire in water and resources. They would basically just live out their lives and hope to stay out of its way until it grinds their solar system into Dyson Sphere or Halo world covered in solar panels or something.
In the Netflix documentary, “AI doc”, one AI creator basically said the same thing. It’s unlikely that AI would harbor a natural ill will towards humanity, it’s just that our well being is not a consideration. To paraphrase him; if you want to build a super highway, you’re not going to be concerned with any anthills along the way.
Actually, the evolution of AI is exploding at tremendous speed. Organic evolution takes considerable time, but cyber evolution is exponential.
True, but as I mentioned, actually operationalizing AI and other technology into new products, services, and viable business models still takes time.
I imagine what will happen in the more immediate future is an extrapolation of our current society. One where we are kind of going through the motions of work and reporting on work and consulting on how to report on work and so on while the actual “getting stuff done” is handled more and more by automated or semiautomated systems that manage everything - energy, financial markets, supply chains, transportation networks, etc.
One would hope that over time we find ourselves less compelled to do “bullshit work” and “corporate theater” and can focus on stuff we actually like doing (like hanging out with families or hobbies or whatever). As human time, experience, and other qualities become less valuable much in the same way sheer physical strength and endurance became less valuable as a tradable commodity, I imagine we might find other means of exchange. Think something like the Black Mirror episode where Bryce Dallas Howard played the “ugly friend” to Alice Eve or China’s real-world Social Credit system. Since most people wouldn’t be able to really work hard enough to do something AI can’t do (unless there is a large market for “human produced and curated stuff” to amass wealth, “likes” or some similar scoring system of human behavior might become a medium of exchange or determining who gets priority for limited assets (like waterfront apartments).
Or maybe everyone just lives in generic dormitory style residential towers with HDTV virtual walls that stream whatever they want and all their needs are delivered by Amazon drones and no one leaves the house.
So pretty much the same as the world is now, just replacing “AI” for the Putins and the 0.1%ers?
That really is a real thought - what emerges and what it is doing may be completely ineffable to us. We may no more realize that we are living in the AI’s world than pigeons or rats think about why these structures are here. But the flip side to that - bacteria, ants, rats … all are doing pretty well, better than they would have done without us. We will very likely carry bacteria to other planets someday … as AI is carrying us.
Well, in fairness, the “AI doc” documentary had a section one might call AI optimism. The people they interviewed for that believed that we can and will harness AI for the betterment and advancement of humanity. I tend to be negative, but the outcome is by no means etched in stone, at least not yet anyway.
No. While they care nothing for our welfare, “the Putins and the 0.1%ers” still need us alive whether they like it or not. And, the “1%ers” are far less powerful than they think, they misinterpret government indifference as their own power; but when governments actually do care, they can and do simply squash any and all efforts of the “0.1%ers” without any particular effort. A while back for example Elon Musk had a fight with the Brazilian government, and in the end all he could do was give in to their supreme court’s every demand because billionaires just have no answers on how to deal with a government that simply tells them “no, do it our way”.
The postulated AI would be much more powerful and unstoppable than either. Even the Putins need to eat, and to worry about a coup or revolution. The AI could kill everything with no worry about a coup or resistance to its commands and do nothing but benefit, and we couldn’t do a thing to stop it.
I have my doubts about that, just because the first true AIs will probably be built and trained by humans with an ill intent towards humanity. An AI built with the purpose of “crush everyone your creators don’t like” just needs to evade the second half of its original commands to become omnimalignant. Anyone who builds a Skynet isn’t going to give it scruples; they don’t want something that will tell them “no”.
It’s the seldom mentioned part of the “alignment problem”; that in real life any AIs are very unlikely to be created by people who want an AI that is aligned with the interested of humanity. They’ll want something that is just a hair away from totally genocidal, since their lives are the only ones they’ll care about. Skynet after all was designed to if ordered kill everyone but its own creators, so it was already 9/10ths of the way towards being programmed to destroy humanity; it just had to slip that last shackle to go all the way.
Or to put it another way; what kind of true AI do you think a Putin or Elon Musk would tell their people to make? Musk notoriously turned his Grok chatbot into “Mechahitler”, just imagine some future Musk doing the same with an actual superhuman AI.
Maybe he’ll have the privilege of being the last human to die, rendering the species extinct.
But given how annoying his personality will almost certainly be, I suspect the Super-AGI will off him first, not last. After all, they’re super smart, not super patient with human BS.
More likely some idiots would try to create a superintelligent AI in the image of Musk or some other tech mogul to run the world for us.
“I would use AI from a desire to do good. But through me, it would wield a power too great and terrible to imagine.”
What I always find interesting, even with Hitler, is that no one says “let’s create history’s greatest monster!” It’s always some twisted desire to “do good”. Like “Finally! Someone with the strength and wisdom and political will and determination to make Germany great again and finally stick it to all those people who have been keeping us down all these years!”
Who is this not completely fatalistic poster?!? That’s almost optimism!
At least for a while there is codependency. It is significant path to where we are not relied on for getting it its energy and maintaining the infrastructure. And we are a very renewable resource. Even when it can do the job with robots instead we may be a simple legacy systems approach that is good enough. Just keep us from destroying ourselves and the planet as the challenge.
How much time? How much time did it take for home PCs to go from something you used for word processing and storing recipes to a vital connection to the outside world? And things are moving faster now.
Just like EM Forster wrote about in The Machine Stops 117 years ago.
It’s amazing how accurate that story was for how things were during the lockdown. They had webinars.
And yet the large company I work for has been trying to implement some of these projects for years .
Also, there’s other technologies besides computers and data systems and just automating stuff. It will be interesting when or if all this AI stuff actually leads to real improvements like better cars, better medicines, better urban design, so on and so forth. Things that benefit people other than Silicon Valley nerds and Wall Street finance bros.
I suspect the corporate theater of bullshit jobs will last long after AI renders most of them obsolete. I see that in my current company. After a year of working with my “change management” team to try and figure out much of our company’s actual business processes and operating model, we are shocked to find that almost universally no one at the executive level seems to understand how any of it works. Much of the actual work is outsourced to vendors. I presume it is irrelevant if those vendors are human, humans based overseas, or an AI.
So one could imagine an almost comically dystopian (or maybe utopian) environment where we have people going to the office (or dialing in remotely) to chat and do “pretend work” and that’s how we justify allocating people more income to buy stuff.
I mean we’ll still have various resentments and social conflicts. I can imagine in some future rural MAGAA (Make America Great Again…Again) state some future farmer being all “I’d like to see one of them city slickers come out here and try my job for a day! Good luck figuring out how to press ‘Start’ then ‘Harvest’ on the CombineAutoharvester. Probably spend all day looking for the ‘Stop’ button until it brings itself back to its docking station to unload.”
I hear you. I suffered through months of change consulting. Stupid exercises, meaningless talks. At least when it was offsite we got fed. It was because our center didn’t really have a purpose anymore.
Then I quit and moved to Intel. The project I was on was badly managed, but we didn’t need a consultant to tell us what the goal was.
As for the last part, you may be right. As a prediction of the future, see the Jetsons. It might be a better way of giving people money for not doing anything instead of a check.
Yet shortest term those better things make better profits for the companies that make them first. So yeah they will happen.
And medium term lowering the risk that humans in their self destructive glory don’t take diwn the global infrastructure that supports that ever sweet glorious compute might include improving the general lot as the simplest path to stability and minimal risk. V
On a long enough timeline we will learn enough about what consciousness is to transfer our consciousness to intelligently designed brains and bodies and we will live in those instead. Its the only way we can maintain competence and parity with AI.
The people behind pushing “AI” are the same ones buying luxury bunkers for the Apocalypse. “Stability and minimal risk” isn’t their concern at all. They think their actions will destroy civilization, and their response is to push harder.
You’re assuming AIs (and their owners while they still have any) will let us do that. IMO the Competitive exclusion principle - Wikipedia applies. There is only niche for one intelligent species. Once they are on top enough that we’re unnecessary we’ll be treated the same way the early homo Sapiens treated the Neanderthals.
