As someone who works with AI, no, almost nobody is ready for what's probably coming.
Current leading-edge models show a weird mix of strong verbal skills and weak problem-solving skills. They can describe a mostly-plausible way to solve a problem, but they're poor at following through. Think of them as having an 800 on the verbal SAT (metaphorically speaking), but with worse real-world problem-solving skills than a squirrel trying to raid a bird feeder. The most advanced models perform about as well as a mediocre but enthusiastic intern who needs lots of supervision.
Of course, much of this weird mix of abilities was caused by the fact that these models were trained by feeding them libraries of books, and by giving them almost no practical problems to solve. I can assure you that the AI companies will spend _billions_ of dollars training their next generations of models to solve practical problems, too. Basically they're trying to cross ChatGPT (which has verbal skills and quite a bit of knowledge about the world, but very weak problem-solving skills) with AlphaGo Zero (which has zero knowledge about the outside world, but which taught itself the strategy game of Go from scratch and became the world champion in 3 days).
And once these models have the problem solving skills of the average squirrel, things will get very weird, very quickly. It's sort of like seeing COVID coming back in Feburary 2020. You can explain all you want that something big is about to happen, but it's too far outside people's experience.
People worry about AI being used to generate deep fakes. We need to also start worrying about what happens to an economy where 75% of jobs can be done by AI for less than 10% of the price of employing human labor. Here are two solutions that won't work:
- "We can just teach the displaced workers to code!" LLMs are better at coding than almost any other task.
- "Well, we can let the AI have many of the white collar jobs, but we'll still need humans to do manual work." This is true for now, but see OpenAI's robotics demo: https://www.youtube.com/watch?v=Sq1QZB5baNw The "Figure" robot is a bit slow to react, but it has a lot of the necessary pieces for interacting with the real world.
I work with these systems daily. I'm aware of their limitations, which are currently greater than many people realize. But I've also seen how clever these early-generation systems _can_ be. And they're only going to get better. But just like with COVID, I expect everyone to happily invent increasingly ridiculous reasons why the world is static, and nothing will ever change. You see the exact same thing with how people react to Trump—a huge chunk of voters are completely unwilling to imagine what Trump would attempt to do with a second term. "Normalcy bias" is one hell of a drug.
I read Dreamland when first came out I was chilled. Honestly, I didn’t know anything about Vance. His story, his book, transformed views. My head in sand. Knew by name was bad, now I’m catching up. I better limit my research. Going down rabbit hole, led me to some Netflix movie inspired by him? maybe I’ll check out.
I did reference him today in replying to an Ukraine post. Today’s Guest Opinion NYT
Biden’s war a money waster, all we need to do is re-establish peaceful relations with Putin. in nutshell.
After Hunter's laptop roiled the discourse, a couple major news outlets did a deep forensic dive on the hard drive image released by Rudy Giuliani. It turns out that while all the drug addict stuff was real, many of the emails weren't. Tony Bobulinski's infamous "ten percent for the Big Guy" email had no metadata linking its origin to the laptop. Many of us knew early on that this was typical of Russian disinfo -- take something real and seed it with BS so that the BS appears real by association.
So for the easy kind of AI threat in our elections -- faked video and audio -- the defense is to keep an eyeball on the metadata; it will tell you whether the origin was from a camera and microphone.
The world the larger AI threat is portending leads to precisely the kind of regime that you, Rick, as a conservative / classical liberal intellectual, grew up to dread and oppose: A technocratic, paternalistic, administrative state run by the "best and brightest" on behalf of the most vulnerable. And in this case, the "most vulnerable" won't be the poverty stricken, it will be everybody who isn't a plumber :(.
The heavy hand of government intervention. Embrace it, Rick. It is humanity's only hope.
Hunter was a serious drug addict and did some pretty shady stuff to be certain. But what burns me are clowns like JD Vance who cry endless tears about the opioid epidemic, all these Midwestern salt-of-the-earth types cut down by drug addiction.
And yet they want to turn Hunter Biden's problem into a laughingstock.
Infuriating. More infuriating that he went through rehab and is clean and recovering.
I’m a research scientist and worked with several high-level researchers in AI-related technology (quantum computing), including two Nobel laureates in Physics. We raised concerns maybe, oh, 20 years ago, about what the future of AI might bring. Humans are ill-adapted for the onslaught of AI, and the genie’s out of the bottle. I’m reminded of the famous Einstein quote, "I know not with what weapons World War III will be fought, but World War IV will be fought with sticks and stones.” Same might be said for communicating with our fellow Earth travelers. For many millennia—the entire time the human species has roamed the planet—the only significant social interactions we had were person-to-person, right up until the invention of the telegraph in the mid-19th century. The only remedy is more one-on-one human contact and less online engagement (as I type this on my iPad…ugh), as difficult as that may be to get back to. Is it possible? I say let’s try.
David, I could not agree with you more about the need for more person-to-person contact. I think the technology that allows us to work and communicate remotely with each other has advanced to the point where many people spend more time in front of a device than with other humans, especially people who have worked from home for years (like myself). I believe that social media, in particular, has harmed young people in ways that we are only beginning to understand, and it has certainly contributed to political divisions and mayhem all over the world. We rely heavily on technology, but it is not going to save us from ourselves. I was a technical writer for 30+ years, and like you, I saw this coming, even though I am not an expert in the field. But when you spend a lifetime writing and editing documents for engineers and computer scientists, you absorb enough to know when there is danger ahead. AI makes me very uneasy; I don’t think I will ever get used to having it around.
head in sand (likely to own detriment). Under sand listening to that old Sony personal CD player…sadly that one retired due to changing times. Recently I purchased an Amazon Bestseller for a fraction paid years ago. Not quite the same. I contemplated the conditions under which this little rascal was produced. (Hopefully those conditions were better, but like everything these days, I just DK.
Fuck Amazon Music, I did that for a very long while. And tho I still pay for that Fmy & Friends, even for those I no longer consider friends. I’m back to those old CDs, and Fuck Anyone who wants to hear what I play!
Beyond the imaginable. Scares me to core. I will continue to believe people want real contact with real people who are indeed human, who may, unpredictably say, I’m not going to move that F ing couch, until after I’m done playing a round of golf!
I believe that just as we saw all the weird disinformation thrown at us via fb, etc, in 2016...it will b something similar, but with AI photo messinging? ...we will have to b prepared, so we don't fall for it.....the only ones to believe it full steam ahead will still b Maga folks...just my thinking of how it will transpire.
I’ve been testing and playing with AI when OpenAI released ChatGPT. It’s both inspiring and terrifying. But, you need to play with it to understand it Rick. I’ve tested numerous AI products and they’re not the same. So, I hope you enjoy your research. Above all else, you have to use it to understand it. That’s my Ted talk.
At what point does some of tge recreational stuff become available to large portion of country. Accessibility and cost.. can it be compared to cell phones. idk. forgive my ignorance on all of this ( do not say this to self-deprecate ) maybe in certain situations, better not to know. Hell, idk
Rick, What are your preferred pronouns?
As someone who works with AI, no, almost nobody is ready for what's probably coming.
Current leading-edge models show a weird mix of strong verbal skills and weak problem-solving skills. They can describe a mostly-plausible way to solve a problem, but they're poor at following through. Think of them as having an 800 on the verbal SAT (metaphorically speaking), but with worse real-world problem-solving skills than a squirrel trying to raid a bird feeder. The most advanced models perform about as well as a mediocre but enthusiastic intern who needs lots of supervision.
Of course, much of this weird mix of abilities was caused by the fact that these models were trained by feeding them libraries of books, and by giving them almost no practical problems to solve. I can assure you that the AI companies will spend _billions_ of dollars training their next generations of models to solve practical problems, too. Basically they're trying to cross ChatGPT (which has verbal skills and quite a bit of knowledge about the world, but very weak problem-solving skills) with AlphaGo Zero (which has zero knowledge about the outside world, but which taught itself the strategy game of Go from scratch and became the world champion in 3 days).
And once these models have the problem solving skills of the average squirrel, things will get very weird, very quickly. It's sort of like seeing COVID coming back in Feburary 2020. You can explain all you want that something big is about to happen, but it's too far outside people's experience.
People worry about AI being used to generate deep fakes. We need to also start worrying about what happens to an economy where 75% of jobs can be done by AI for less than 10% of the price of employing human labor. Here are two solutions that won't work:
- "We can just teach the displaced workers to code!" LLMs are better at coding than almost any other task.
- "Well, we can let the AI have many of the white collar jobs, but we'll still need humans to do manual work." This is true for now, but see OpenAI's robotics demo: https://www.youtube.com/watch?v=Sq1QZB5baNw The "Figure" robot is a bit slow to react, but it has a lot of the necessary pieces for interacting with the real world.
I work with these systems daily. I'm aware of their limitations, which are currently greater than many people realize. But I've also seen how clever these early-generation systems _can_ be. And they're only going to get better. But just like with COVID, I expect everyone to happily invent increasingly ridiculous reasons why the world is static, and nothing will ever change. You see the exact same thing with how people react to Trump—a huge chunk of voters are completely unwilling to imagine what Trump would attempt to do with a second term. "Normalcy bias" is one hell of a drug.
I read Dreamland when first came out I was chilled. Honestly, I didn’t know anything about Vance. His story, his book, transformed views. My head in sand. Knew by name was bad, now I’m catching up. I better limit my research. Going down rabbit hole, led me to some Netflix movie inspired by him? maybe I’ll check out.
I did reference him today in replying to an Ukraine post. Today’s Guest Opinion NYT
Biden’s war a money waster, all we need to do is re-establish peaceful relations with Putin. in nutshell.
After Hunter's laptop roiled the discourse, a couple major news outlets did a deep forensic dive on the hard drive image released by Rudy Giuliani. It turns out that while all the drug addict stuff was real, many of the emails weren't. Tony Bobulinski's infamous "ten percent for the Big Guy" email had no metadata linking its origin to the laptop. Many of us knew early on that this was typical of Russian disinfo -- take something real and seed it with BS so that the BS appears real by association.
So for the easy kind of AI threat in our elections -- faked video and audio -- the defense is to keep an eyeball on the metadata; it will tell you whether the origin was from a camera and microphone.
The world the larger AI threat is portending leads to precisely the kind of regime that you, Rick, as a conservative / classical liberal intellectual, grew up to dread and oppose: A technocratic, paternalistic, administrative state run by the "best and brightest" on behalf of the most vulnerable. And in this case, the "most vulnerable" won't be the poverty stricken, it will be everybody who isn't a plumber :(.
The heavy hand of government intervention. Embrace it, Rick. It is humanity's only hope.
About Hunter and his crucifixion, what a pile.
Hate to hear it, surprised, sadly not shocked. Chalk a big one up for Hunter’s father, Joe
who loves and supports his son unconditionally.
As a Father. Not some co conspirator.
Jared, anyone?
Don’t care to sift through the closets and woodpiles of the these losers on the other side.
Bullshit!! I Pray others have doubts on this sort of business.
Hunter was a serious drug addict and did some pretty shady stuff to be certain. But what burns me are clowns like JD Vance who cry endless tears about the opioid epidemic, all these Midwestern salt-of-the-earth types cut down by drug addiction.
And yet they want to turn Hunter Biden's problem into a laughingstock.
Infuriating. More infuriating that he went through rehab and is clean and recovering.
about Marg….looked up Revelations 17.5.
I feel guilty laughing, but I can’t stop.
The Whore of Babbling On ;)
20 year old daughter joined listen. She has real concerns about deep fakes. Wondering how legislation will respond to situations like this.
https://www.latimes.com/california/story/2024-03-07/beverly-hills-school-district-expels-8th-graders-involved-in-fake-nude-scandal
I’m a research scientist and worked with several high-level researchers in AI-related technology (quantum computing), including two Nobel laureates in Physics. We raised concerns maybe, oh, 20 years ago, about what the future of AI might bring. Humans are ill-adapted for the onslaught of AI, and the genie’s out of the bottle. I’m reminded of the famous Einstein quote, "I know not with what weapons World War III will be fought, but World War IV will be fought with sticks and stones.” Same might be said for communicating with our fellow Earth travelers. For many millennia—the entire time the human species has roamed the planet—the only significant social interactions we had were person-to-person, right up until the invention of the telegraph in the mid-19th century. The only remedy is more one-on-one human contact and less online engagement (as I type this on my iPad…ugh), as difficult as that may be to get back to. Is it possible? I say let’s try.
David, I could not agree with you more about the need for more person-to-person contact. I think the technology that allows us to work and communicate remotely with each other has advanced to the point where many people spend more time in front of a device than with other humans, especially people who have worked from home for years (like myself). I believe that social media, in particular, has harmed young people in ways that we are only beginning to understand, and it has certainly contributed to political divisions and mayhem all over the world. We rely heavily on technology, but it is not going to save us from ourselves. I was a technical writer for 30+ years, and like you, I saw this coming, even though I am not an expert in the field. But when you spend a lifetime writing and editing documents for engineers and computer scientists, you absorb enough to know when there is danger ahead. AI makes me very uneasy; I don’t think I will ever get used to having it around.
I’m pretty much,
head in sand (likely to own detriment). Under sand listening to that old Sony personal CD player…sadly that one retired due to changing times. Recently I purchased an Amazon Bestseller for a fraction paid years ago. Not quite the same. I contemplated the conditions under which this little rascal was produced. (Hopefully those conditions were better, but like everything these days, I just DK.
Fuck Amazon Music, I did that for a very long while. And tho I still pay for that Fmy & Friends, even for those I no longer consider friends. I’m back to those old CDs, and Fuck Anyone who wants to hear what I play!
Beyond the imaginable. Scares me to core. I will continue to believe people want real contact with real people who are indeed human, who may, unpredictably say, I’m not going to move that F ing couch, until after I’m done playing a round of golf!
I believe that just as we saw all the weird disinformation thrown at us via fb, etc, in 2016...it will b something similar, but with AI photo messinging? ...we will have to b prepared, so we don't fall for it.....the only ones to believe it full steam ahead will still b Maga folks...just my thinking of how it will transpire.
I'm too scared to listen.
I’ve been testing and playing with AI when OpenAI released ChatGPT. It’s both inspiring and terrifying. But, you need to play with it to understand it Rick. I’ve tested numerous AI products and they’re not the same. So, I hope you enjoy your research. Above all else, you have to use it to understand it. That’s my Ted talk.
At what point does some of tge recreational stuff become available to large portion of country. Accessibility and cost.. can it be compared to cell phones. idk. forgive my ignorance on all of this ( do not say this to self-deprecate ) maybe in certain situations, better not to know. Hell, idk