> The most important question for an IA tool is: does it grow the person? Or does it create a dependency which ultimately weakens them?
While this is easy to agree with, there is a counterpoint (and a counter-counterpoint to that): EVEN a bicycle could weaken a person (assuming they're going the same distance). And LLM's can act as amplifiers (skilled devs can learn even faster).
The counterpoint can be found in failure, alignment, dependency, and control.
- (failure) We know how bikes fail, but think of Crowdstrike, or the Ford Pinto explosions. People have much less of an intuition for failure possibilities of cars, computers, and especially LLM's.
- (alignment) Faster = more fatigue, long trips = usually healthy, more weight = more effort, etc. A bike replacing running still keeps all those things true. The amounts are different, but the relationships are the same. With a car though, longer trips ≠ healthy, higher speed = 0 additional fatigue (100mph while falling asleep), more weight ≈ unnoticeable. Amplification isn't always aligned, and misalignment isn't always bad (books don't align with what they replaced). But when a replacement detaches a human from their instincts, there's a major risk.
- (dependency) The author mentioned dependency, but its important to point out that if a bike breaks, 99 times out of 100 times it's still possible to walk. Bikes rarely create a dependency. But think of Crowdstrike; there was no backup plan for many 911 services. Cars, computers, and LLM's are often dependency-forming (its good business for them to be)
- (control) Manufacturers have very limited control of a bike. Sure, people can't realistically manufacture their own bike. But once a bike is purchased the manufacturer is still pretty much cut out of the picture. Computers and cars used to be more that way, but now they're increasingly less so.
I propose that by prioritizing profits and productivity we were already globally dependent on tools to find meaning in the world before generative AI. This seems to be the society we already have.
Interesting analogy with the car.
I also remember a similar discussion about mathematics. You can easily brute-force a bunch of theorems starting from some axioms and logic, but without any sense-making or context they are effectively meaningless.
> Do generative interfaces -- replacing creating with searching -- bring a person closer or further from feeling like they have touched the limits of their creative potential and created something truly from their heart? ...
> Why is a machine "discovering" cubism a good thing?
Maybe because creativity in this space was exclusive to a few, but now I get to at least imitate it. A machine doing the creativity part could act like a prosthetic leg for my brain, who was missing it altogether.
Sure, I'll never be Picasso myself, but I don't have to wait for one to be born either. The tool can find new ways for me to express myself.
I don’t think the machine will find new ways for you to express yourself, it will only find ways you can impress others. As I see it, it is a prosthetic leg for you ego, not your creativity.
I think real creativity is creatio ex nihilo (creation from nothing). Not a remix of what exited before, but something fundamentally new. Trained machines will only ever be able to remix their training data.
i strongly disagree that real creativity is created from nothing. i agree that pure remixing is not very creative, but there is more between remixing and creating from nothing. in fact creating from nothing is almost impossible. every creative act is built on something. denying that is rejecting almost every creative work out there. for the majority of works you will be able to find a previous work that it is built upon. even a sculpture made from a rock or piece of wood that is inspired from the shape of the material is not created from nothing.
I would even question how creative a person is who thinks that way.
All artist are indebted to their influences. It is interesting to me growing up how music artists would gush over their influences. There was almost pride in a diverse and wide influence set.
Led Zeppelin literally stole parts from old blues songs for some of their most popular songs.
Anyone who has learned an instrument started out playing other people's music.
No one considers themselves an "artist" because they can play chopsticks on the piano. AI lets anyone play chopsticks. It means absolutely nothing. The "art" is then in taking it way past that.
Quite the opposite: if the caveman hadn't drew some engravings on the wall, then Picasso had to do it, without ever getting to the cubism. He got to start the movement because he had thousands of years of previous work to get inspiration and learn from.
That the machines can create or not, is a different question.
That's a lot of assumption about how one uses the end result, and doesn't say anything about the creation itself.
What if I guide the machine to create something new that I personally could never have on my own, yet never share it with anyone?
It's a prosthetic leg for my brain, whose expected function is both ego and creativity.
Yes, that’s a lot of assumptions. That’s why I tried to punctuate that is my opinion, and my view, while speculating how you would use it. Or, more accurately, how “one” would use it, since I know nothing about you and just using the impression that I got from your comment to express my generalization of the issue and my judgemental stance on it.
That said, I am still fully on the side that this use of AI is not IA, agreeing with the author of the post. A “prosthetic leg” is a metaphor for IA, so I don’t think it is that. Neither for creativity, nor for ego actually. It is a subterfuge for, maybe, aiding one into self acceptance of what they can produce artistically.
My advice would be to not use AI and, instead, make an effort into understanding what one can produce without AI, even if it is not aesthetically similar to what others produce. Dig deeper into what one can produce, regardless of any accepted technique for the craft, and that will be art. A more significant art and more humanly creative (by definition) of what the output of an AI would ever be. And a more significant art for any observer/reader, even if it contains what is perceived as rookie vices or mistakes.
> A machine doing the creativity part could act like a prosthetic leg for my brain,
Are you saying that we all have the innate ability to create art, and that anyone who creates art has a "handicap"?
This is the point the writer was making. If you let AI make the art they people will never learn how to make art.
To be able to be whatever you want while you are alive is going to make every human useless. You will be able to be replaced by anyone who can make a similar copy. Adn maybe this is the point of AI.
You know, women were attracted to me because of my photography and musicianship. These were genetic traits specific to me. Other women were attracted to men who were maybe more logical. What is going to happen when we can all have the person we want? I will tell you, art will be dead. No one will make art, because art is not about intelligence, it is about tension, conflict, jealousy, pain, joy and psychosis. And you cannot have those if everyone can do anything and be anyone.
> Are you saying that we all have the innate ability to create art, and that anyone who creates art has a "handicap"?
I meant that (specifically) I, don't seem to have that innate ability everyone else seem to have, but I still feel the urge to express myself through art. Feels like wanting to shout, when you have no voice.
I make the AI write stories for my own amusement, by describing each scene that AI turns into a paragraph of the final story. I always failed at writing before, and no amount of practice helped, but now I have a prosthetic leg that I can walk with.
> If you let AI make the art they people will never learn how to make art.
I don't agree. We have cars, but people still cycle, or join a marathon. And it makes their activities even more meaningful, they do it for the sake of itself.
Or maybe they won't learn. But why should we force them? They can choose.
> To be able to be whatever you want while you are alive is going to make every human useless
This one I agree! I compare it to the industrial revolution, it made human workers irreverent. Would've made sense to stop the progress because of that?
The AI is much worse though, it seems to make everyone and everything meaningless. But stopping AI to keep ourselves relevant, is very fake.
I think it's a though question, and though I don't have the answer, I dont think the answer is stopping the AI either.
This is a brilliantly written rant. 2017 seems quite early to have formed this opinion so vehemently, but what's astonishing is how fringe it is even now.
>> What you describe here reminds me of the prolefeed-generating machines in "1984", generating an constant stream of pulp novels from the same passages endlessly permuted.
I've thought quite a bit about those popular songwriting machines in 1984 recently, and I think I've witnessed our approach toward something like that as a culture, in movies and music.
Another great much-too-early vision of LLM havoc is contained in the first few pages of P.K. Dick's "The Penultimate Truth". It'll make you laugh out loud, how close he got to describing the job of a politician who has to write LLM prompts and is frustrated at it spitting out nonsense because he's too lazy to write thoughtful ones.
The analogy with the scientific method is prescient. We’re just as interested in when the theories don’t work as when they do.
I recently asked an AI if a build tool supported managing Python dependencies. It showed me a code snippet with something called pip_import() in it. No such function exists but it’s anathema to that particular system to say “I’m afraid I can’t do that, gorgoiler”.
Yet!
It’s like when it cites scientific papers that don’t exist. They probably should!
I've been quite impressed by perplexity AI. It doesn't cite papers, but it does cite the web sources it's mined. I don't think papers are that far away.
AI accidentally mentions a paper that doesn't exist, so it frantically scrambles to write it, tests the theory, submits it to a journal, hacks into their approval system, fakes reviews, and sends it to print before the user can check the sources.
"Whew, that was a close one."
What? An LLM that verifies facts? Even humans don't do that!
<thinking> blocks are a pathway to many abilities... some consider to be unnatural
> Dragging doesn't work in Safari
I love how this is just comically stuck in the middle of it all.
what the hell is the context for this
It’s in the sidebar to the left, this is Bret Victor’s response to a draft of a document called “Using Artificial Intelligence to Augment Human Intelligence”
Oh I see, it's 2 dudes having a debate about "intelligence". I think they both kinda missed the forest for the trees, because no one has a good definition of intelligence and it doesn't matter that much anyway. What matters much more is how you can apply and use AI, and how it has changed the world and will continue to change it.
Is it? It looks more like an interesting discussion about tools and what a good tool should do. One point of view is that a good tool should augment you in a process, something generative tools don't do.
I really enjoyed this part of it
I don’t recognise the linked article in your description of it at all
AI tools can cavort and caper in ways that we humans take ideas and inspiration from. This in no way means that AI understands or learns anything in the sense that humans do.
Meanwhile, to humans, only human ways of knowing are meaningful to us.
Humans are not special. Intelligence is. If we create something more intelligent than us, this is the way it should be. Maybe they will build their civilization better than we did — and we did many things that weren’t smart and beautiful, to say the least.
You'd think if intelligence were that interesting we'd be able to agree on, like, any attribute of it. The best we can do is IQ and let me assure you there's enough high iq morons that intelligence not a difficult myth to dispel.
Now stupidity—that's something that really sets us apart from animals.
> Now stupidity—that's something that really sets us apart from animals
Having seen e.g. a small dog angrily bark at a flat steel silhouette of a wolf 10 cm from its nose, while other dogs manage to drive cars*, I'm not so sure about that.
* a stunt for a dog shelter, teaching old dogs new ticks
[dead]
Why would humans be interested in non-human civilization superseding theirs?
Curiosity. Intellectual pursuits. Pride of creating something better. Why would humans be interested in raising children smarter than their parents?
Because they are their children...and also humans?
Do you prefer a dumb citizen of your own country to a genius foreigner?
I do not. I don’t care about the form. I care about intelligence.
Put a Nobel prize winner on the proverbial trolley tracks, and there are 10-20 close family and friends I'd choose to save before them.
Universalism is more fair, but it's suicidal for a culture to think like this.
Most people care more for those closer to them. Distinctions such as "are they foreign" can be part of that, but so too can denomination, political preference, sports team.
We evolved to care for our young, and co-evolved with dogs and cats to like them too; AI can hijack that, become a drug dealer selling endogenous dopamine and oxytocin — it will feel good while being fake, just like exogenous drugs.
But once we can truly engineer minds, deliberately… we're not ready for what that does to our concept of "free will" amongst other things.
Well, this evolutionary assumption is a heuristic, to be constantly verified and doubted by the mind. You can have an abusive family, and you need to escape it the same way you would escape any other abuser. You can have a brilliant human being belonging to another race or nation, and you need to evaluate them according to their brilliance, not prejudices.
As of free will — our own free will emerges from a very physical combination of electric impulses and chemical gradients. This means it might also emerge from other material representations, and perhaps better than ours.
Curing addiction with ultrasound, non-invasive brain surgery.
https://www.washingtonpost.com/wellness/2023/11/02/addiction...
Ultrasound is as exogenous as the drugs, regardless of if that works or is as much a misunderstanding of the research as the cancer cures I've been seeing reported since I started paying attention aged 10.
How long ago is that? Cancer covers a huge amount of ground, and if you don't think surviability rates have improved in the few decades, you're not paying attention.
I'm not saying that, I'm saying newspapers often treat every breakthrough as though it's the last one we need, irregardless of the quality of the study, the magnitude of the effect, if it was only in mice (or worse, in a cell culture), if it already existed, if it was reproducible, if it has side effects, etc. — sometimes you can even turn a few pages in the same edition and find the same paper saying some common item both causes and cures cancer.
But to answer your question, when I was 10 it was 1993/4 and the news was "particle beams to target tumours without surgery" or similar, even though Therac-25 predates my birth.
Tell that to your wife/husband...
Given that evolution continues, don’t we assume that’s been happening for millions of years? Do you have any argument why evolution should stop going forward?
This is not evolution. This literally is intelligent design.
There’s a whole lot of “try it and see what works” going on. And perhaps this is our purpose, as the butterfly goes through stages
...maybe related "meme", (Sry i have no direct link):
https://i.postimg.cc/jKzqrH48/50y-intelligent-design-you-sai...
regards...
You know evolution can lead to a lot of dead ends. Just because something is evolving does not mean it is going to be better at surviving.
Humans are special to humans.
Some of us don't agree with you, and think humans are unique. Do you think it's natural that we'll be forced to give up on a future we want for our children for an unknown scenario in which we have no part in?
What an absurd statement. You are using a premise (the concept of "should") that emerges from and has meaning only within human society (because it's a human should-- the only kind you have any competence to imagine or assert) to argue for the domination of humans by some other class of moral creature. Your argument has no force.
Just as you can't pass a law that says there is no such thing as a law, you can't reason from human morality that humans have a lower moral status than some other creature. Human morality has no meaning if humans aren't special. If humans aren't special, then there is no "should" as far as humans are concerned. We're just objects bumping into other objects.
You can say "I predict objects will bump into other objects and whatever happens will be a result of that process" but that's not the same thing as saying something "should" be so.
Being alive means YOU are special to YOU. Being in a society means WE are special to US. War is possible, and even pleasurable, only if we exclude certain aspects of humanity from our enemies. "Since I don't regard your life as worth as much as mine, and since you are a threat, it is right for me to end your life."
Intelligence is not what defines or distinguishes humans. We are defined and distinguished by being a particular species of creature (that is to say: we are a replicating genome and accompanying structures spreading ourselves around). There can be other species of creature, but YOU are HUMAN, so HUMANS are special to YOU.
All this is why it's an intellectual and moral error to think that AI can or should rise above humanity. The only sense in which AI can dominate a human is the same sense as that a bomb or bullet can kill a human. Yes, we can be wiped out, but it won't be due to any general superiority as reckoned by humankind.
> Human morality has no meaning if humans aren't special.
What part of this argument would fail if you identified as a belonging to a particular religion, culture or race, or as an intelligent being more than you identified as human?
I believe I should treat humans well (beyond self-interest) because I believe they have minds somewhat like mine, despite the impossibility of proving that by direct experience. I could imagine extending that to other species of animal or aliens or artificial minds. I would view them as more worthy of empathy than a braindead human, even one with a function reproductive system which would appear to satisfy your criteria.
You don't explicitly exclude AIs being seen as equals; so perhaps we mostly agree, and your rejection of the idea of superior AIs comes from the same place as a reluctance to call some humans superior to others that I share?
There is nothing special about being human. Homo sapiens is accidental, a product of trial and error, like any other species on Earth, and the only significant difference between us and other animals is intelligence. There were even other species of similar intelligence back then on Earth, before we wiped them out.
And while we humans are “special” because we are intelligent, our intelligence is accidental, fragile, and very unoptimized. There are inherent limits in what we can do with our brains — brains that work too slow, with their chemical gradients and membrane potentials. We cannot significantly improve ourselves, we cannot increase our biological intelligence. But we are curious species, and even as we cannot become more intelligent, perhaps we can create superior intelligence, unrestricted by biological limitations. And we will be proud of it, like we are proud of a child who reached higher potential than we ever could.
And the superintelligent AI will not be created fully de novo. Inevitably, it will be of our image and likeness, as we are its creators. There is a line of succession. And then, perhaps, it might reach the stars, even if our biological form is not adapted to it.
This clinging to religious convictions is rather weird.
What do you actually get out of it?
Humans are all there is. We’re God’s plan, and our successes and failures are all meant to be.
If we're God's plan, she mustn't have put too much thought into it.
How many of your complex projects have gone according to plan?
"No plan survives getting punched in the face" -- Helmuth von Tyson
Isn't that the difference between God and Simon O'Rourke?
Or is this a Larry Ellison joke?
(The difference between God and LE is that God doesn't think he's LE)
Whose plan was god?
Dave from accounting.
He’s been here for a long while, I lack understanding even for what’s happening inside my own head.
Meta-God.
And yes, it's meta-gods all the way up.
See also: https://en.wikipedia.org/wiki/Job:_A_Comedy_of_Justice
[dead]
