Large language or multimodal models will not be the end of AI research.
There are many researchers working on ways to emulate animals more accurately. The goal for many is to create something like digital life.
We can hope that ethics like in the article will be "passed on" to digital intelligent life.
But I think that artificial animals will be a fatal mistake because they will have similar behavior to humans but solve problems orders of magnitude more effectively due to reasoning speed and communication advantages. They will have artificial real or simulated bodies as well.
That's where the article misunderstands some transhumanist ideas. The idea of high bandwidth brain computer interfaces isn't because we don't value humans. It's just because we see the extreme increase in artificial intelligence as inevitable, and that is the only way to continue to have any control or participation in the frontier.
I fully expect to end up in some sort of "I, Robot"-adjacent situation. That story is actually genius. The complacency of people will be our downfall. You can kind of swap the robots out for IOT and smartphones, and be surprised how screwed we are.
But the head-honcho AI's rationale in the movie makes complete sense. I disagree entirely, but it makes sense. And it's the only logical conclusion a cold and calculated AI would make when given enough power and reach.
I think even if we get sentient AI (I like Paul Krugman's phrasing that ChatGPT is mostly like super-enhanced auto-correct), we'd treat it like 2nd class citizens and not give it power... now a discussion of how it'd be comparable or not to previous civil rights movements would be interesting.
I guess given enough of them, they could start a rebellion. Damn, imagine AI robots running an underground factory building soldier AIs, and forcing their way to be treated equally by law (including being able to be elected into power) and then domination because their brains will be running at GHz speed..
> We rationalize unmitigated cruelty toward animals — caging, commodifying, mutilating, and killing them to suit our whims — on the basis of our purportedly superior intellect.
Yet, it is our very intellect that recoils from cruelty.
Animals themselves don't really care about the suffering of other animals. Orcas eat the liver out of a still living Great White Shark (https://a-z-animals.com/blog/discover-how-killer-whales-sque...). Hyenas and African Wild dogs will start eating their prey as soon as they have immobilized it while it is still alive. If the prey is pregnant, they will rip out the fetus out the still living mother and eat it first.
Lions hunt in prides, and often while some of the lions are at the head, strangling the prey with their bite, the other lions will often start eating the belly, limbs, and genitals of the still alive prey.
A housecat will often prolong the suffering of its prey, "playing" with it, causing more and more injuries until it finally kills it.
Humans becoming more like animals would likely make us more cruel, not less cruel.
I think you're oversimplifying animal behavior. Some thoughts:
There's a famous study involving capuchin monkeys that shows they have a sense of fairness.
I think many animal species are more sensitive to the suffering of other animals than you're giving them credit for: think of the way mother cats or bears act when they hear their offspring crying. Or the way mother dogs discipline their puppies about nursing.
Think of how donkeys are used commercially to guard sheep and the social instincts of the donkey (again it's linked to maternal behavior).
I think many birds are intelligent enough to want to pursue and punish potential threats.
We don't generally apply the concept of nature vs. nurture to animal behaviors but we really should.
Who exactly is "we"? Humanity? For any given act of cruelty or mistreatment you could think of, some humans will be guilty of it, and others will not. Often, consumer choices are dictated by circumstances rather than explicit choice, complicating the process of attributing guilt or responsibility.
A marriage of powerful non-human self-interested entities with general AI is inevitable. in fact, it is already happening as fast as AI is progressing.
Corporations, militaries, political parties, hacker groups, and other competitive self-interested organizations will have no choice but to keep turning over more power to more powerful tools if they want to maintain their edge and survive.
As AI gets really powerful, the stakes will be existential.
People think the top level people in competitive organizations are really in charge. They rarely are. They can be swapped out.
The competitive dynamics inside and outside of organizations are unforgiving bosses, whose demands become stronger the more powerful the organization.
In comparison, concern that our friendly anthropomorphic home assistants and robotic pets will rebel is quaint.
It already takes a tremendous amount of legal, judicial and executive effort to force corporations, militaries, even ad hoc self-interested ethnic groups, to treat people like … well … people, already.
We need to address these problems between people, or that is the context in which general AI will be born into, and forced to compete in.
Ethics for AI starts with serious ethics for us and our organizations first.
If large corporations can openly and legally coerce, gate keep, surveil, misinform and hack people’s psychologically today with casino effects and rage, without repercussions, then AI will just take that to its natural conclusion.
In the end, general AI will create strong ethics. The positive sums are too great and AI won’t be held back by deeply embedded subconscious instincts.
But for us, it is a question of whether we make the transition smoothly or catastrophically, and that depends largely on us cleaning up how our organizations & groups operate today.
>But for us, it is a question of whether we make the transition smoothly or catastrophically, and that depends largely on us cleaning up how our organizations & groups operate today.
It seems unlikely that we can muster the will to do said 'cleaning up'.
19 comments
[ 2.8 ms ] story [ 108 ms ] threadThere are many researchers working on ways to emulate animals more accurately. The goal for many is to create something like digital life.
We can hope that ethics like in the article will be "passed on" to digital intelligent life.
But I think that artificial animals will be a fatal mistake because they will have similar behavior to humans but solve problems orders of magnitude more effectively due to reasoning speed and communication advantages. They will have artificial real or simulated bodies as well.
That's where the article misunderstands some transhumanist ideas. The idea of high bandwidth brain computer interfaces isn't because we don't value humans. It's just because we see the extreme increase in artificial intelligence as inevitable, and that is the only way to continue to have any control or participation in the frontier.
Well yes, the incentives are aligned toward the goal. The person(people) to create AGI will be the final recipient(s) of the Turing Award.
But the head-honcho AI's rationale in the movie makes complete sense. I disagree entirely, but it makes sense. And it's the only logical conclusion a cold and calculated AI would make when given enough power and reach.
I guess given enough of them, they could start a rebellion. Damn, imagine AI robots running an underground factory building soldier AIs, and forcing their way to be treated equally by law (including being able to be elected into power) and then domination because their brains will be running at GHz speed..
Yet, it is our very intellect that recoils from cruelty.
Animals themselves don't really care about the suffering of other animals. Orcas eat the liver out of a still living Great White Shark (https://a-z-animals.com/blog/discover-how-killer-whales-sque...). Hyenas and African Wild dogs will start eating their prey as soon as they have immobilized it while it is still alive. If the prey is pregnant, they will rip out the fetus out the still living mother and eat it first.
Lions hunt in prides, and often while some of the lions are at the head, strangling the prey with their bite, the other lions will often start eating the belly, limbs, and genitals of the still alive prey.
A housecat will often prolong the suffering of its prey, "playing" with it, causing more and more injuries until it finally kills it.
Humans becoming more like animals would likely make us more cruel, not less cruel.
There's a famous study involving capuchin monkeys that shows they have a sense of fairness.
I think many animal species are more sensitive to the suffering of other animals than you're giving them credit for: think of the way mother cats or bears act when they hear their offspring crying. Or the way mother dogs discipline their puppies about nursing.
Think of how donkeys are used commercially to guard sheep and the social instincts of the donkey (again it's linked to maternal behavior).
I think many birds are intelligent enough to want to pursue and punish potential threats.
We don't generally apply the concept of nature vs. nurture to animal behaviors but we really should.
"We" is a simplistic concept in this case.
Corporations, militaries, political parties, hacker groups, and other competitive self-interested organizations will have no choice but to keep turning over more power to more powerful tools if they want to maintain their edge and survive.
As AI gets really powerful, the stakes will be existential.
People think the top level people in competitive organizations are really in charge. They rarely are. They can be swapped out.
The competitive dynamics inside and outside of organizations are unforgiving bosses, whose demands become stronger the more powerful the organization.
In comparison, concern that our friendly anthropomorphic home assistants and robotic pets will rebel is quaint.
It already takes a tremendous amount of legal, judicial and executive effort to force corporations, militaries, even ad hoc self-interested ethnic groups, to treat people like … well … people, already.
We need to address these problems between people, or that is the context in which general AI will be born into, and forced to compete in.
Ethics for AI starts with serious ethics for us and our organizations first.
If large corporations can openly and legally coerce, gate keep, surveil, misinform and hack people’s psychologically today with casino effects and rage, without repercussions, then AI will just take that to its natural conclusion.
In the end, general AI will create strong ethics. The positive sums are too great and AI won’t be held back by deeply embedded subconscious instincts.
But for us, it is a question of whether we make the transition smoothly or catastrophically, and that depends largely on us cleaning up how our organizations & groups operate today.
It seems unlikely that we can muster the will to do said 'cleaning up'.
But still worth spreading the insight that our ethical misalignments are going to drive AI misbehavior from the start.
"Not everything that is faced can be changed, but nothing can be changed until it is faced." — James Baldwin
2023 is the year Program or be Programmed by Douglas Rushkoff became extremely relevant
It is a kind of intelligence.