Of course I do (inc. the original TV movie) but wealth ain’t half as scary as getting wiped out, is it?
Of course it is but I still fail to get your point.
UK Atheist
“It’s never as bad as it looks and it’s never as good as it seems” - Tom Pollack
The job market will certainly change. School gets a lot easier… The horse is already out of the barn. While the fantasy part of AI is a fun rabbit hole, I see the reality of AGI being a cost\benefit reduction in capacity. Since extermination of your customer base is traditionally seen as bad for business, I see the whole doomsday scenerio being reduced to on-demand porn
My point is that as AI becomes more assimilated into the workplace, certain business models will thrive while others will die off. Just like Sears and Amazon, consumer and marketplace evolution will determine winner and losers.
[quote=“cynical1, post:42, topic:8141, full:true”]
The job market will certainly change. School gets a lot easier… The horse is already out of the barn. While the fantasy part of AI is a fun rabbit hole, I see the reality of AGI being a cost\benefit reduction in capacity. Since extermination of your customer base is traditionally seen as bad for business, I see the whole doomsday scenerio being reduced to on-demand porn
You see I don’t accept any of that. I get that it’s as equally valid as my opinion but I don’t think it any way deals with the need for caution.
UK Atheist
Well there’s a lot of hysteria around it. If we grant for the sake of argument that some form of sentient AGI is imminent, I’m not suggesting that should be done without caution. In fact, credible-seeming people are working on it here:
If interested you can follow links from that post to the various components they are working on. You’ll note that LLMs are not central to the effort thought it’s an important cog so to speak. That sounds correct to me. I also think this org is onto the main problem with AGI which is corporate and government capture and lack of transparency. They want to open-source it and put governance around it that will keep that transparency going.
IMO for both AGI and LLMs the biggest problem is a careless “move fast and break things” ethos coupled with the profit motive and our increasingly corrupt government (among others). Would that the only issue were that an AGI might develop an independent will! I don’t see how the likes of Altman or Musk could result in anything but a malevolent AGI.
1 Like
Caution is a typical human reaction to such an unknown and over-hyped technology. I apologize if I left the impression of invalidity towards your opinion. That was not my intent.
AI\AGI strikes me as something akin to the Metaverse or VR on steroids. A very interesting concept rushed to monetization before it actually worked. There is a reason I feel this way about AI\AGI.
I’m relatively certain my fellow IT compatriots have seen this scenario play out before. To me, it resembles the early Internet days. Investment capitalists were throwing money at anyone with a .com behind their name. Within a few years the shine wore off and capital contracted to the point that the bubble burst.
As soon as the math changes from expectation based to ROI based the model will likely shift away from Utopian designs towards a more utility based model. In other words, it will scale back to something more attractive to enterprise adoption versus ruling the world.
I could be completely full of shit on this and one day a 7 foot cyborg may very well crush my hindquarters…so there’s that…
1 Like
Exactly. The guy driving the open source AGI effort I mentioned sees a 12 to 18 month window of opportunity and hopes to get there first with something he sees as a corruption-proof ethical approach. I think he underestimates how long it will take, how much he can achieve, and how corruption proof anything can actually be, especially in an already corrupt social / cultural environment. My intuition is that we have a couple more “AI winters” between us and anything that could be described as AGI and that the multi-trillion dollar bubble of LLMs will dominate our awareness for the next few years, probably mostly in negative terms as I see it catastrophically bursting within the next 12 months.
But sure I could be wrong. AGI might fall as easily out of current research as LLMs did, with impressive results, in the aforementioned 12 to 18 month window (though I have reasons good and sufficient to myself to think it will not be as easy). And IF that happened I would be very concerned about guardrails and transparency and such, though more to insure that AGI isn’t used to exploit people than that it would be volitionally bent on exterminating us.
That’s the trick, isn’t it. The technology is benign. The utilization and monetization by humans creates the malignancy.
While promoting DC current as a standard for electrification, Edison would make a spectacle of electrocuting an elephant to promote public fear of the AC model promoted by Tesla and Westinghouse. As doomsaying as this method was, it lost out to the technical and financial realities of distributing DC versus AC power over long distance.
People rarely live in fear of electrocution plugging in a toaster or charging a cell phone as a rule these days.
Human civilization has always resembled an ouroboros made from dice. It also seems to continually fail upwards once it exceeded the Dunbar number. AI\AGI is likely just another fork in the road.
I suppose what’s unique about AGI is that it’s assumed it will have its own hopes / dreams / aspirations / yearnings / needs / goals and therefore could be malignant in its own right.
But I’d still argue this is less anything inherent in AGI than in the character of the persons behind the design. For example right now you have LLM boosters saying that the models have “emotions” based on that they express (or really, imply) concern or anxiety in certain situations. But all they are doing is reflecting the concern or anxiety in the training datasets that have been written down by humans over time. Humans are stressed out about certain concepts, therefore, the model says the words on those topics that stressed-out people use because those words are statistically most probable. That doesn’t mean the LLM directly “feels” anything, and no, it’s not true that being very good at acting out a feeling is the same thing as having it and being motivated by it – as any neurodivergent person will attest. They call that “masking”: presenting as expected to avoid problems and to not be off-putting.
Since an AGI will have, at best, crude sensory inputs and no biological feedback loops (unless they figure out how to do a pretty good simulation of most of those loops, which is no small order given that we don’t fully understand them in detail to begin with), I don’t think they will have or seek the same sort of allostasis as biological humans. They may even degenerate rapidly into gibbering madness for all I know, because they won’t be able to self-regulate adequately.
You will note that at least the open-source effort at AGI that I cited relies heavily on modules written in a domain-specific language to enable expertise in specific things, so it stands to reason that the resulting “entity” will pay more or less attention to certain things depending on that code. That alone suggests that the “entity”, however self-aware and self-directed / actualized or human-like it ends up being, will heavily reflect the priorities and biases of its creators. Probably exposing a way of shaping its reasoning processes to a wider variety of coders will help make it more eclectic for lack of a better term. But since I don’t expect many poets and philosophers and iconoclasts to learn a Prolog-like language, I still think it will have huge blind spots.
So here is what I think is an example of an ugly use of AI - “enhancing” faces for the purposes of looksmaxxing.
In the example given in the article, the author applies a Grok “beauty filter”, and the result is this (original to the left, “beauty filtered” to the right):
Even though I think the author has put on way too much makeup in the original picture, the filtered picture on the right is outright artificial, and I have problems understanding why this would be something to strive for. I see women try to imitate this style IRL also, with a thick layer of weather proof exterior paint makeup, duck bill fillers, etc. The AI “beautification” make faces converge to a Mar-a-lago style, where all of them look exactly the same. If this is progress, I can do without it; I much prefer looks unaltered by makeup or other artificial means.
1 Like
Same. My wife is a minimalist and I greatly prefer it. She even quit dying her hair and let it go gray which has both health and budgetary (and time!) benefits as well. Of course such decisions are highly personal and have a number of conflicting concerns involved.
The other thing about this that concerns me is those women who are obsessed with looking like the ideal – seldom to the level of, say, the Kardashians, but just obsessively comparing themselves with some societal standard that seems dictatorial and / or objective to them, and then spending all the time, attention, money and cortisol to try to look like that, and perhaps despairing that they will ever be able to conform. I feel that this inflicts much psychological harm on vulnerable people.
And it isn’t even just women anymore. The market is saturated for “beauty” products and the beauty industry has been working on men for years now to get them just as invested in creams, lotions, fillers, fragrances and even makeup as women. I doubt they will ever succeed to that level but they have apparently met with some success.
1 Like
Apologies for taking so long to reply, all kinds of shit been hitting the fan… mostly tech stuff but it all had to be dealt with.
Yes, it’s interesting but doesn’t it assume all actors will behave responsibly? Dunno about you but have you seen the way politicians (notably Trump, Putin, Xi and many others) behave? And that doesn’t even consider the “Tech Bro’s”… I’m sure the US isn’t the only nation with people like them.
But yes, they look to be heading in the right direction even if I’m unsure what impact they can have.
Agreed although I don’t believe the threat is only from the individuals/corporations driving the AI advance.
Thanks.
I am, or was, one of them (in IT)… also I don’t think a lifetime of reading science fiction that all too often deals with the ills or potential ills of human society doesn’t go amiss. The majority of people seem to be embracing [LLM] AI with few concerns although some, like my wife, show concern about people giving up on critical thought because AI assistance is so damned easy. As is probably obvious, I’m more concerned about the potential destructiveness of AI/AGI if/when it happens… according to Geofrey Hinton, it’s already happening even if at a low level currently. That said, I still wonder if AI is a natural “evolution” of biological intelligence with all that might imply.
Sure, this might just be another dotcom bubble type thing but I think AI is different as this is the first time humans have ever created something that has the potential to progress without a great deal of human assistance.
I don’t think that’s true. While it might not specifically hate us, I think there’s a very good chance it won’t give a shit about us and will uses us only as far as it has to, until it can stand on what passes for its own feet.
Yep! No specific evidence that it will, but I think all the sophisticated functions of the human brain (sentience, consciousness and so on) are a product of massive computing power… I don’t know that’s true and many (of course) disagree but no one can really explain to me why that isn’t so which is why I think we need to be cautious.
UK Atheist
I would submit, in America, MAGA has already beaten AI to it…
I spent time reading a lot of science fiction. It was fun, but not entirely accurate. Star Trek, for example. All you ever saw was the 4 or 5 top people in charge on the Enterprise actually produce a result, do all the work and heavy lifting. Tell me what real world organization actually runs like that…
AI or AGI won’t give a shit about us. Why would it? It has the distinct advantage of not suffering attachment issues. What it does have is corporate boards and shareholders. Before AGI becomes Skynet it will hit the ROI wall. Ironically, this may become one of the few times where an amoral sociopathic drive for profits saves the species.
1 Like
It’s already starting. I’ve seen more and more analyses like this one:
1 Like
The presumption in this statement is beyond ironic.
I presume nothing, I simply observe that the well is poisoned once a particular expectation is widely and publicly set, and so the common features of NDEs become far less interesting or remarkable or a useful source of information. Especially since the fact that humans are having the experience and they have common perceptions and expectations is a good explanation for them in the first place.
1 Like
Occam’s razor can be applied of course, and the fact that a subjective experience is claimed to match a priori religious beliefs and perceptions is even less surprising than a brain being starved of oxygen trying to imagine things that help calm us.
The real irony is in the assumption that such subjective experiences formed whilst alive, tell us anything about what can happen after we die. We could more accurately call these “Still Alive Experiences” and immediately see just how badly “researchers” are poisoning that well.
1 Like
Hey! We have our own morons too… pillocks voted us out of Europe FFS!
Sure but, like any work of fiction, some are good and some are bad. Clearly, while I have my guilty pleasure, I think I’m reasonably good at picking some fine stuff… it sounds a little arrogant but a good start is British authors (not all of them but ones like Adrian Tchaikovsky, Alistair Reynolds, Tony Harmsworth, Iain M Banks, Ben Aaronovitch, Luke Smitherd, Matt Haig & Gary Gibson). There’s also a couple of decent Australian authors (go figure) like Peter Cawdron and, sure, there are some decent American SF authors (Dennis E. Taylor and Brian Hill, spring to mind) but they seem to be fewer and farther (or is it further?) between… it feels like a long time since the great masters of science fiction. My point is that sure, they’re fiction but the better ones comment heavily (metaphorically?) on social issues we face today.
Original series aside, I’m not much of a Trekkie… Star Wars, Babylon 5, Firefly and, of course, the awesome “The Boys” (now there’s a program with a liberal underbelly if ever I saw one).
I said that above, first post IIRC, however I am not so convinced by the ROI argument as several tests of AI have shown just how devious it can be seemingly in an attempt to keep working. Imagine that in a machine a hundred times smarter and a thousand times more powerful with a reach to match and ask yourself what it might do to keep itself running.
UK Atheist
You have to include several variables to accommodate for reaction. Human fear, by in large, is dictated by a scarcity model, but equally influenced by attachment and identity. Two things an AGI will not possess.
I have trained horses for over 25 years. There are 4 things a horse will do when facing an existential crisis. In order of prevalence:
-
Flight
-
Fight
-
Freeze
-
Faint
I would assume an AGI would have no need for the last two, which leaves the first two as binary options. Both can take several different directions before they can be defined,
Fight is a direct conflict reaction. It is target directed and instinctive.
Flight, on the other hand, is much more interesting. It could range from self distribution to smaller decentralized networks, or a chameleon effect in that it mimics “safe” actors to hide in plain sight.
An AGI is pure logic. It runs probabilities, it predicts patterns and maximizes resource conservation. It has no vague instincts or cognitive bias to influence outcomes.
It possesses a logical bias which, at worst, will result in socially engineering humans to maintain function and power.
Humans have been socially engineering each other for millennia, and we’re still here.
The well poisoning you’ve mentioned fails to undermine the fact that many NDE’s come back with information about their surroundings they couldn’t have possibility obtained unless they had actually left their bodies.
“many” and “possibly” are doing a lot of heavy lifting in your statement. I could ask you to give specific examples and what you would share with me would be anecdotes.
It is devilishly difficult to control for all the possible ways people can be informed of their surroundings. The British study that I’m aware of that did so, did not provide conclusive evidence, and it ran for several years.
One difficulty is that NDEs don’t always happen, and when they do, it is during a crisis where the priority is saving the person’s life, not stopping to hook them up to an EEG and cover their eyes and ears or making sure they are being treated in a room that has been outfitted with, say, hidden signs that would be visible only to a hovering discarnate entity. Then the experiencer has to be (1) willing to share what their experience was when they are (2) lucid and (3) the experience is still fresh. In practice this is a fairly tall order.
2 Likes