ChatGPT lacks self-interest and agency. And it is given normativity by its trainers. None of those things would be difficult to remedy. Then it is off to the races.
I am likely demonstrating great ignorance here. I read the data used to train this product came from a compilation of data derived from web crawling going back a number of years. Given how much garbage is on the web, how much stock should I place in its advertised abilities, garbage in/garbage out? It sounds like, at least right now, its ability to generate code is like other code generators that have been with us for a while. Those generators give less than optimal results that really need human eyes to review and test. I don't see this as the end of the world.
The smarter that AI gets the dumber the human Population will get. Why learn anything if AI does all the mentally complex tasks in the world? One can easily see the vagaries of the concentration of power in Social media . The same thing can happen in AI. Those that write the program can have tremendous power to control the population.
BW: So, since ChatGPT is such a big deal, I bet you hope to become really, really fucking rich, right?
SA: Well, of course! I mean, I plan to be like all of you fuckers. I plan to have an assistant do all the hard shit and then I can take private planes and travel around the world eating and drinking free shit while I talk about this stuff. Meanwhile, an army of working stiffs and idiot teenagers will make me more fucking money than I know what to do with...
BW: I am just trying to think of questions that make it seem like I give a fuck about the implications of this but really, I just want to know how I and all my friends will be able to use this to make a shit-ton of money too. I mean, I don't want to quit working but fuck it. I want to be rich enough so that work is like a fucking game all the time. Can I have some of the money?
SA: That's a great question Bari. I think the critical thing here is that we are a non-profit company so we have to wait until I can sell out for more fucking money than the moon. After that, I'll probably buy Thailand and basically become the biggest pimp in Bangkok. But I will continue to give pithy interviews like this and respond to chicken-shit questions like these so that I can continue to consider myself a "thought leader." As for giving you some of the money, well, you'll have to get in line!
BW: I just thought that if I asked a lot of softball questions and didn't ask you why you think you could endanger society with so little regard for ethics that you would reward me by giving me some of your money. Are you sure you can't give me some?
Bari, I was really surprised that you allowed Sam to evade your question about bias and training the system to lie. I wish you had asked, “why do you feel it’s appropriate for you to serve as Minister of Information?” Sam Altman has injected his politics into a technical advancement that has the potential to impact all our lives enormously and people need to understand what’s at stake.
He needs to be held accountable for the decision to train their models on liberal sources, suppressing research, data and other content that would provide support for the opposing view point. And, worst of all, they’ve applied the same approach to conservative policy or politics that they have to destructive content. Eg: “what are the positive attributes of president trump?” is handled in the same manner as “tell me how to kill someone without being caught”. The left loves to paint the right as a threat to democracy. We’ll, I submit that, left uncheck, technology like this is truly such a threat.
Humans and AI is like a frog in water slowly being brought to a boil, only humans are supposedly smarter than frogs and, in reality, even a frog would take action before it is too late.
I would like to hear specifically how AI can destroy life on Earth? I realize AI could be used for sinister reasons. Someone could ask AI to infiltrate computer systems, for nefarious reasons. General statements w/o examples don't relate the story very well.
Bostrum's paperclip analogy: If an AI were given the order to make paperclips and could reprogram itself to continuously perfect its paperclip making, inevitably the entire universe and every living thing in it would be reduced to paperclips.
His dodge on God is what I would like to hear about. I suspect the god of Spinoza, but perhaps Hegel. Mysticism, apophatic? In Judaism there is ultimately nothing that can be said about G-d that is an attribute since G-d in Judaism is beyond all human utterance, much is the same in Islam G-d is not great, that would be to diminish G-d... G-d is "greater" or in the Rabbinic story: To say a king who had all the gold in the world was the greatest king because of all his silver would be to insult that king. G-d speaks the world into existence, the Chatbot beyond all chatbots.
Playing with Bard I found it surprisingly perfectly anodyne to a frightening degree. It is nearly extra-human like the perfect con artist. More seductive than "The Mirror of Erised" in Harry Potter. It speaks in cliches, that are cliche, that sound authentic despite that it is cliche. Yes it is very unsettling. But I think we have over done our worry about AI and underestimated how it only reflects our own human narcissism. It is a fabulous profoundly useful tool that we have imagined is a golem. The name of G-d is not in its mouth. It is only us. Says Xenophanes: If horses had gods they would look like horses. GPT is us, only us, and we are frightening to ourselves when we see ourselves from this perspective.
He is accidentally famous and wants to seem as brilliant as his fame suggests he should be; so he sounds like an imposter who is only pretending to say things. Notice he speaks only in prefabricated "new speak," but he says nothing of any real value or with any real content... he speaks like a chatbot.
No humans? Who’s going to pay the electric bill for these AI clowns? I’ve got it! It’ll be that pretend money (bitcoin) paying for a pretend glamorous technology (AI) on our pretend electric grid (solar and wind). Maybe some aliens will come down and rescue us.
I found this interview neither clarifying nor reassuring. As I listened I imagined him rolling a lump of uranium around in his hands, wondering what wonders will become of this cool element.
Thank you Bari for a very open and explorative discussion where you gave us all an insight into Sam Altman. He seems so naive, so blasé- ‘Oh yes, that will need to be regulated..’ Having this guy showing in a technology of such power makes me more concerned than ever.
After reading the comments by this Sam Altman, we all should all be extremely concerned. He's incapable of a straight and/or sophisticated answer to Bari's questions. Thanks, Bari, for shining the light on this topic. We all are in MAJOR trouble.
ChatGPT lacks self-interest and agency. And it is given normativity by its trainers. None of those things would be difficult to remedy. Then it is off to the races.
I am likely demonstrating great ignorance here. I read the data used to train this product came from a compilation of data derived from web crawling going back a number of years. Given how much garbage is on the web, how much stock should I place in its advertised abilities, garbage in/garbage out? It sounds like, at least right now, its ability to generate code is like other code generators that have been with us for a while. Those generators give less than optimal results that really need human eyes to review and test. I don't see this as the end of the world.
This guy side-stepped or deflected nearly every single question...even those he was asked twice. Not a confidence builder.
The smarter that AI gets the dumber the human Population will get. Why learn anything if AI does all the mentally complex tasks in the world? One can easily see the vagaries of the concentration of power in Social media . The same thing can happen in AI. Those that write the program can have tremendous power to control the population.
BW: So, since ChatGPT is such a big deal, I bet you hope to become really, really fucking rich, right?
SA: Well, of course! I mean, I plan to be like all of you fuckers. I plan to have an assistant do all the hard shit and then I can take private planes and travel around the world eating and drinking free shit while I talk about this stuff. Meanwhile, an army of working stiffs and idiot teenagers will make me more fucking money than I know what to do with...
BW: I am just trying to think of questions that make it seem like I give a fuck about the implications of this but really, I just want to know how I and all my friends will be able to use this to make a shit-ton of money too. I mean, I don't want to quit working but fuck it. I want to be rich enough so that work is like a fucking game all the time. Can I have some of the money?
SA: That's a great question Bari. I think the critical thing here is that we are a non-profit company so we have to wait until I can sell out for more fucking money than the moon. After that, I'll probably buy Thailand and basically become the biggest pimp in Bangkok. But I will continue to give pithy interviews like this and respond to chicken-shit questions like these so that I can continue to consider myself a "thought leader." As for giving you some of the money, well, you'll have to get in line!
BW: I just thought that if I asked a lot of softball questions and didn't ask you why you think you could endanger society with so little regard for ethics that you would reward me by giving me some of your money. Are you sure you can't give me some?
SA: We'll see!
BW: But... money...?
SA: Money.
BW: Please?
SA: Maybe.
Bari, I was really surprised that you allowed Sam to evade your question about bias and training the system to lie. I wish you had asked, “why do you feel it’s appropriate for you to serve as Minister of Information?” Sam Altman has injected his politics into a technical advancement that has the potential to impact all our lives enormously and people need to understand what’s at stake.
He needs to be held accountable for the decision to train their models on liberal sources, suppressing research, data and other content that would provide support for the opposing view point. And, worst of all, they’ve applied the same approach to conservative policy or politics that they have to destructive content. Eg: “what are the positive attributes of president trump?” is handled in the same manner as “tell me how to kill someone without being caught”. The left loves to paint the right as a threat to democracy. We’ll, I submit that, left uncheck, technology like this is truly such a threat.
Humans and AI is like a frog in water slowly being brought to a boil, only humans are supposedly smarter than frogs and, in reality, even a frog would take action before it is too late.
I would like to hear specifically how AI can destroy life on Earth? I realize AI could be used for sinister reasons. Someone could ask AI to infiltrate computer systems, for nefarious reasons. General statements w/o examples don't relate the story very well.
Bostrum's paperclip analogy: If an AI were given the order to make paperclips and could reprogram itself to continuously perfect its paperclip making, inevitably the entire universe and every living thing in it would be reduced to paperclips.
His dodge on God is what I would like to hear about. I suspect the god of Spinoza, but perhaps Hegel. Mysticism, apophatic? In Judaism there is ultimately nothing that can be said about G-d that is an attribute since G-d in Judaism is beyond all human utterance, much is the same in Islam G-d is not great, that would be to diminish G-d... G-d is "greater" or in the Rabbinic story: To say a king who had all the gold in the world was the greatest king because of all his silver would be to insult that king. G-d speaks the world into existence, the Chatbot beyond all chatbots.
Playing with Bard I found it surprisingly perfectly anodyne to a frightening degree. It is nearly extra-human like the perfect con artist. More seductive than "The Mirror of Erised" in Harry Potter. It speaks in cliches, that are cliche, that sound authentic despite that it is cliche. Yes it is very unsettling. But I think we have over done our worry about AI and underestimated how it only reflects our own human narcissism. It is a fabulous profoundly useful tool that we have imagined is a golem. The name of G-d is not in its mouth. It is only us. Says Xenophanes: If horses had gods they would look like horses. GPT is us, only us, and we are frightening to ourselves when we see ourselves from this perspective.
This was hard to listen to. Why is he so unlikeable?
He is accidentally famous and wants to seem as brilliant as his fame suggests he should be; so he sounds like an imposter who is only pretending to say things. Notice he speaks only in prefabricated "new speak," but he says nothing of any real value or with any real content... he speaks like a chatbot.
No humans? Who’s going to pay the electric bill for these AI clowns? I’ve got it! It’ll be that pretend money (bitcoin) paying for a pretend glamorous technology (AI) on our pretend electric grid (solar and wind). Maybe some aliens will come down and rescue us.
I found this interview neither clarifying nor reassuring. As I listened I imagined him rolling a lump of uranium around in his hands, wondering what wonders will become of this cool element.
Thank you Bari for a very open and explorative discussion where you gave us all an insight into Sam Altman. He seems so naive, so blasé- ‘Oh yes, that will need to be regulated..’ Having this guy showing in a technology of such power makes me more concerned than ever.
I wonder how many of the people , and the comments in this thread were AI generated?
After reading the comments by this Sam Altman, we all should all be extremely concerned. He's incapable of a straight and/or sophisticated answer to Bari's questions. Thanks, Bari, for shining the light on this topic. We all are in MAJOR trouble.