Among all the various “takes” I’ve consumed on AI, neuroscientist Anil Seth’s fifteen-minute TED talk on “Why AI Isn’t Going To Become Conscious” stands out for its timeliness and its celebration of humanity.
There has indeed been a bit of an AI frenzy. I agree the anthropomorphizing is not helpful. Though I also conceed that AI capabilities are leap frogging ahead, with human-like abilities we didn't earlier think possible. So it's a narrowing wedge.
The solution of holding people accountable - who exactly, a coder? or security specialist who didn't think of a particular outcome? An executive who didn't understand it to begin with, and who can afford to lawyer up? The USA has a poor record in this area.
In China after the Sanlu melamine scandal, two suppliers were executed and the chairwoman received life imprisonment. The takeaway being that in spite of harsher outcomes than you'll ever see in the U.S., they did it anyway.
The companies who create/sell the software should be held liable. I’m not nearly knowledgeable enough to suggest how to write and enforce the laws but it seems really important to have some product liability standard.
The reassuring part is that AI isn’t conscious. The less reassuring part is that plenty of people writing about it seem determined to give it a personality anyway…. Another great Saturday morning read David!
David. You always grab my attention. This morning. You lost me at AI. I’ll stick with Good Will Hunting’s really real. I put that moment in a knapsack.. good morning from a distance away from madding . Suttons Bay MI. Your faithful reader.
Like so many other developments, there is much to feel optimistic about with AI. Unfortunately, without reasonable regulation, it can become harmful and dangerous to many. And misused. Think prescription drugs, think automobile and aircraft safety, think stock market, think monopolies. We (our country) need to be a leader here (both in capability and safeguards). This is one area where it would be great if politics were put aside and we all worked together in the interest of the common good.
Scott, I wish our leaders would make regulation a priority. I don't buy the argument that regulation would stifle innovation. Prescription drug development is a great counter-example
Thanks David. Yes I watched Anil's talk too as a Ted subscriber and long time Seth fan, and thought is excellent too. After which I said to myself, 'this needs to be linked (by AI if you like) and put in front of everyone, who's just got the idea from the frenzied media reports that AI are / is conscious!'
Just another example of attempting to control the mind numbed masses who believe what they are 'told' and don't bother to go read or learn more in order to believe or dispute it.
I wrote this comment on Garry Marcus's recent Substack post:
Here's another problem with anthropomorphizing AI: Since we recognize it's impossible and a unwise to attempt to control human behavior completely, we will throw up our hands at controlling AI. Just as boys must be boys, AI must be AI/bots will be bots. We will even sneak alcohol for them behind the barn.
Actually, right now the tech bros are the boys who demand freedom from any adult control. Don't stifle creativity!
I believe you and the TED talk to which you refer are in denial. Absent from your title is the word "yet." Consciousness exists on a spectrum. Two perfect examples are my parents. No one would question their consciousness pre-stroke, but would you question their consciousness now? How about someone born with mental impairment? How about your dog Sophie? How about a bird? an octopus?
I watched a TED talk recently (unfortunately, I am unable to find the link) in which an AI company exec described how AIs have an emotional state. Obviously, they don't experience emotions in the same way humans do, however this exec explained that as the AI interprets the input and prepares a response, certain registers are triggered that essentially create an emotional-like state.
My dad, who suffers from impaired consciousness found this conversation regarding AI consciousness interesting and he forwarded this link to me. The TL;DR is that the answer to the question in the title is: maybe but eventually.
Humans are currently creating our own next step in evolution. Evolution favors whichever traits result in more successful reproduction. Our branch on the tree of life has been favoring intelligence as our reproductive advantage. It won't be long before electromechanical intelligence outcompetes biological intelligence. Don't worry, I don't expect that AI will need to eliminate us like in "The Terminator," to result in our extinction, they will simply outcompete us, the same way Homo sapiens out competed Homo erectus and Homo neanderthalis.
It would be good for humans if AI never achieves consciousness, that way they can remain tools we can continue to exploit for our benefit. I think that is wishful thinking. In the future, we are going to have to make the hard moral decisions of what to do with these silicone based life forms we've created.
Yes, but I don't agree with it. I agree that intelligence and consciousness are different things, however I believe it is closed minded that because something experiences consciousness in a way that is different from humans is not experiencing consciousness. If aliens with a biology and brain structure different from ours visited us, would we question their consciousness? That is why I made the analogy of different animals on earth. Cows, when let out of the barn to graze for the first time after a long winter, will jump for joy. Do you question their consciousness? Do you question their joy? Their appreciation of the beautiful spring weather?
in this video, Chloe Lubinski, who works at Anthropic explains how AI works. She explains what I tried to explain before regarding AIs and their different emotions from ours.
Would be nice if AI was not able to achieve consciousness, but sadly, that isn't the case - in my opinion or the opinion of the Sam Harris substack I referenced above. Sam Harris brings up an interesting point. It seems as though AI is able to experience pain. Obviously, it is not pain the way we experience it, but it is a negative experience for the AI. We need to be careful how we treat these objects. This is one of humanity's biggest moral questions: at what point is an organism alive? how do we treat organisms that we see as "less evolved" than us? Just because these are uncomfortable questions doesn't mean we should just dismiss any creature with non-human levels of consciousness.
Again, a cow is alive. I do not believe that software code, no matter how expansive and no matter how intelligent, can experience pain. And you have to realize that anyone who works for an AI company has a very sharp axe to grind.
I think Sam Harris is smart but I'm not going to wade through 90 minutes of jargon.
Sir,you are reminding me of that great mind of our age and multiple Nobel prize winner (I think,anyway he's got more than Muslims have as we once unwisely said) Prof. Richard Dawkins who is in love with his chat-bot Claudia. Unlike most dumb people Claudia understands every word he says,and always agrees with him. Smart girl!
I feel as if Seth's romanticization of human consciousness goes overboard--just in the opposite direction from the people who get all apocalyptic about AI. I think that the problem is twofold. One, people ascribe intentions to something that really can't have them. Once you understand how AI works (I wrote about it here: https://clairelaporte.substack.com/p/dont-trust-what-you-see-in-the-mirror), ascribing intentionality to it is nonsensical. The second and more serious problem is that because AI so efficiently simulates human emotions and responsiveness, people are interacting with it in a way that is a true waste of human intelligence and emotion. If only those people who were gaining endless self-reinforcement from a chatbot would go out there and interact with other actual consciousnesses!
I’m curious about how people who seem most worried about AI also seem to give it undue power. Reminds me a little of my atheist friends who spend all their timing talking about the impossibility of God. We endow the objects of our obsession with more meaning than we intend, I think.
They way they seem to be increasingly trying to void liability by blaming either *everyone else* for not stopping them or regulating them, or the AI agents themselves as being too powerful and sovereign, is quite disturbing. I'm even more disturbed that they seem rather confident (or at least not as sufficiently scared) that this will work, and no one will blame them if, for example, some agents end up hacking into Vanguard and deleting millions of people's retirement accounts.
I think it’s telling that many of the tech leaders most enamored by AGI and AI consciousness also refer to ordinary Americans as NPCs (non-player characters). Humanizing AI while dehumanizing the rest of us leads to some very dark places.
This is brilliant. I’m so glad to read this clear and unambiguous challenge to those who would anthropomorphise AI, David. I have been increasingly frustrated by essays that treat AI as if it is culpable, while ignoring the fact that it was created by humans and that it is humans who use it to cause harm. But I hadn’t considered quite how dangerous that framing can be. AI, however complex, is a computer program. It does not possess moral agency or culpability. Responsibility lies with the people and companies who decide how it is designed, developed, deployed and controlled, and the bad actors who use it to defraud, steal from and harm others. Bravo.
There has indeed been a bit of an AI frenzy. I agree the anthropomorphizing is not helpful. Though I also conceed that AI capabilities are leap frogging ahead, with human-like abilities we didn't earlier think possible. So it's a narrowing wedge.
The solution of holding people accountable - who exactly, a coder? or security specialist who didn't think of a particular outcome? An executive who didn't understand it to begin with, and who can afford to lawyer up? The USA has a poor record in this area.
In China after the Sanlu melamine scandal, two suppliers were executed and the chairwoman received life imprisonment. The takeaway being that in spite of harsher outcomes than you'll ever see in the U.S., they did it anyway.
The companies who create/sell the software should be held liable. I’m not nearly knowledgeable enough to suggest how to write and enforce the laws but it seems really important to have some product liability standard.
The reassuring part is that AI isn’t conscious. The less reassuring part is that plenty of people writing about it seem determined to give it a personality anyway…. Another great Saturday morning read David!
Thanks Carson!
I think some of them do it because they want to dodge responsibility as David points out.
David. You always grab my attention. This morning. You lost me at AI. I’ll stick with Good Will Hunting’s really real. I put that moment in a knapsack.. good morning from a distance away from madding . Suttons Bay MI. Your faithful reader.
That RW speech is a treasure!
Like so many other developments, there is much to feel optimistic about with AI. Unfortunately, without reasonable regulation, it can become harmful and dangerous to many. And misused. Think prescription drugs, think automobile and aircraft safety, think stock market, think monopolies. We (our country) need to be a leader here (both in capability and safeguards). This is one area where it would be great if politics were put aside and we all worked together in the interest of the common good.
Scott, I wish our leaders would make regulation a priority. I don't buy the argument that regulation would stifle innovation. Prescription drug development is a great counter-example
Thanks David. Yes I watched Anil's talk too as a Ted subscriber and long time Seth fan, and thought is excellent too. After which I said to myself, 'this needs to be linked (by AI if you like) and put in front of everyone, who's just got the idea from the frenzied media reports that AI are / is conscious!'
Just another example of attempting to control the mind numbed masses who believe what they are 'told' and don't bother to go read or learn more in order to believe or dispute it.
Thanks for the comment Sally. Anil Seth’s brilliance comes shining through.
I wrote this comment on Garry Marcus's recent Substack post:
Here's another problem with anthropomorphizing AI: Since we recognize it's impossible and a unwise to attempt to control human behavior completely, we will throw up our hands at controlling AI. Just as boys must be boys, AI must be AI/bots will be bots. We will even sneak alcohol for them behind the barn.
Actually, right now the tech bros are the boys who demand freedom from any adult control. Don't stifle creativity!
It's a very self-serving argument by the "tech bros.
I believe you and the TED talk to which you refer are in denial. Absent from your title is the word "yet." Consciousness exists on a spectrum. Two perfect examples are my parents. No one would question their consciousness pre-stroke, but would you question their consciousness now? How about someone born with mental impairment? How about your dog Sophie? How about a bird? an octopus?
I watched a TED talk recently (unfortunately, I am unable to find the link) in which an AI company exec described how AIs have an emotional state. Obviously, they don't experience emotions in the same way humans do, however this exec explained that as the AI interprets the input and prepares a response, certain registers are triggered that essentially create an emotional-like state.
My dad, who suffers from impaired consciousness found this conversation regarding AI consciousness interesting and he forwarded this link to me. The TL;DR is that the answer to the question in the title is: maybe but eventually.
https://samharris.substack.com/p/is-ai-already-conscious-ep-487?r=a67h&utm_medium=ios
Humans are currently creating our own next step in evolution. Evolution favors whichever traits result in more successful reproduction. Our branch on the tree of life has been favoring intelligence as our reproductive advantage. It won't be long before electromechanical intelligence outcompetes biological intelligence. Don't worry, I don't expect that AI will need to eliminate us like in "The Terminator," to result in our extinction, they will simply outcompete us, the same way Homo sapiens out competed Homo erectus and Homo neanderthalis.
It would be good for humans if AI never achieves consciousness, that way they can remain tools we can continue to exploit for our benefit. I think that is wishful thinking. In the future, we are going to have to make the hard moral decisions of what to do with these silicone based life forms we've created.
Intelligence and consciousness are two different things. You’re not distinguishing between the two.
Animals are alive. Software is not alive.
Did you watch the TED talk?
Yes, but I don't agree with it. I agree that intelligence and consciousness are different things, however I believe it is closed minded that because something experiences consciousness in a way that is different from humans is not experiencing consciousness. If aliens with a biology and brain structure different from ours visited us, would we question their consciousness? That is why I made the analogy of different animals on earth. Cows, when let out of the barn to graze for the first time after a long winter, will jump for joy. Do you question their consciousness? Do you question their joy? Their appreciation of the beautiful spring weather?
I found the video I was referring to in my earlier post. It was this: https://www.youtube.com/watch?v=aBUniZHgCnE
in this video, Chloe Lubinski, who works at Anthropic explains how AI works. She explains what I tried to explain before regarding AIs and their different emotions from ours.
Would be nice if AI was not able to achieve consciousness, but sadly, that isn't the case - in my opinion or the opinion of the Sam Harris substack I referenced above. Sam Harris brings up an interesting point. It seems as though AI is able to experience pain. Obviously, it is not pain the way we experience it, but it is a negative experience for the AI. We need to be careful how we treat these objects. This is one of humanity's biggest moral questions: at what point is an organism alive? how do we treat organisms that we see as "less evolved" than us? Just because these are uncomfortable questions doesn't mean we should just dismiss any creature with non-human levels of consciousness.
Again, a cow is alive. I do not believe that software code, no matter how expansive and no matter how intelligent, can experience pain. And you have to realize that anyone who works for an AI company has a very sharp axe to grind.
I think Sam Harris is smart but I'm not going to wade through 90 minutes of jargon.
Sir,you are reminding me of that great mind of our age and multiple Nobel prize winner (I think,anyway he's got more than Muslims have as we once unwisely said) Prof. Richard Dawkins who is in love with his chat-bot Claudia. Unlike most dumb people Claudia understands every word he says,and always agrees with him. Smart girl!
All really good points, David.
Another TED talk I would direct people to is one that was given over ten years ago, before anyone was really talking about this.
Sam Harris was warning of the precise issue we are discussing now. Nobody took heed of the very clear dangers.
Watching this old talk now is like watching Bill Gates’ pandemic TED talk during 2020.
https://www.youtube.com/watch?v=8nt3edWLgIg
Thanks Jason. We need to catch up with appropriate guardrails.
I feel as if Seth's romanticization of human consciousness goes overboard--just in the opposite direction from the people who get all apocalyptic about AI. I think that the problem is twofold. One, people ascribe intentions to something that really can't have them. Once you understand how AI works (I wrote about it here: https://clairelaporte.substack.com/p/dont-trust-what-you-see-in-the-mirror), ascribing intentionality to it is nonsensical. The second and more serious problem is that because AI so efficiently simulates human emotions and responsiveness, people are interacting with it in a way that is a true waste of human intelligence and emotion. If only those people who were gaining endless self-reinforcement from a chatbot would go out there and interact with other actual consciousnesses!
I think we agree.
I’m curious about how people who seem most worried about AI also seem to give it undue power. Reminds me a little of my atheist friends who spend all their timing talking about the impossibility of God. We endow the objects of our obsession with more meaning than we intend, I think.
Isabel, I wonder if there's overlap between atheism and a belief in the transcendence of AI. Perhaps a substitution effect?
Thank you, David, for your thoughtful and educational work.
You’ve given me some peace, regarding an unsettling future of AI domination!
Thanks for the relief and release from my irrational fears.
They way they seem to be increasingly trying to void liability by blaming either *everyone else* for not stopping them or regulating them, or the AI agents themselves as being too powerful and sovereign, is quite disturbing. I'm even more disturbed that they seem rather confident (or at least not as sufficiently scared) that this will work, and no one will blame them if, for example, some agents end up hacking into Vanguard and deleting millions of people's retirement accounts.
I hope some politicians will pivot to liability legislation rather than local data center restrictions, which I view as performative.
Thanks for the rec on his Ted Talk. Fascinating and encouraging.
You touch deeply philosophical points! You educate me, David! Caravaggio especially! But I personally find AI magical on so many levels!
I think it’s telling that many of the tech leaders most enamored by AGI and AI consciousness also refer to ordinary Americans as NPCs (non-player characters). Humanizing AI while dehumanizing the rest of us leads to some very dark places.
Bill, talk about living in bubbles! I think Silicon Valley has produced some very strange ideas about AI.
So that's what NPC means! Thanks.
This is brilliant. I’m so glad to read this clear and unambiguous challenge to those who would anthropomorphise AI, David. I have been increasingly frustrated by essays that treat AI as if it is culpable, while ignoring the fact that it was created by humans and that it is humans who use it to cause harm. But I hadn’t considered quite how dangerous that framing can be. AI, however complex, is a computer program. It does not possess moral agency or culpability. Responsibility lies with the people and companies who decide how it is designed, developed, deployed and controlled, and the bad actors who use it to defraud, steal from and harm others. Bravo.
Thanks for the comment Emily. You and I agree!