Glenn, I'm more of a skeptic on this than you seem to be. So far I have avoided using any of it, as much as I can. When it was starting up, I talked to my older son about it. He just turned 50, and he's been working as a computer programmer for about 30 years. No degree--we got him his first computer when he was 8, after we realized he was talking to his friends on the phone--telling them what to type into their Texas Instruments home computer, and they'd tell him what happened. By the time he was 12, I would take him to a Radio Shack Users' Group meeting, and the guys in their 50s would be asking him to explain what was going on inside their computers. His take on AI was that it is mostly hype.
I have already been seeing reports of copyright lawsuits filed against the AI companies, because their bots (and their programmers) are accessing copyrighted material online without asking permission.
But I have one more issue with this stuff, and it might be considered philosophical: yes, these AI entities can access information online. The problem I see--and this is often a long-standing problem for humans as well--there is a serious difference between "knowledge" and "wisdom." Even with humans, academic "knowledge" is no guarantee of "wisdom." We can see that in a lot of the professors showing up at protests all over the country. And I have known some very wise men in 75 years of life who didn't have much in the way of academic credentials. This past week I have been reading some books and diaries by Eric Hoffer--a farm laborer and longshoreman who ended up a professor at Berkeley. (He had a low opinion of most intellectuals!) One case of this we have been facing is this "gain-of-function" research, which has done a lot of damage world-wide in the last few years. "Knowledge" may tell you how to do it; but whether you should do it or not is a matter of "wisdom"--and I am afraid the last few years are evidence that our scientists and their government supervisors were lacking in wisdom.
Glen: I’m not sure why you added the paragraphs downplaying the importance of intelligence. But since you did: I couldn’t disagree more.
There *is* an evolutionary arms race in intelligence. Intelligence has arisen along at least three completely independent paths: mammals, birds, and octopi. It’s so important that Mother Nature invented it three different times.
Told my wife a few hours ago how much I love Grok. I’ve used it this weekend for tips on how to grill pork chops, to verify sources of dietary fiber, ask which state has the most species of raptors (CA followed closely by Texas), ask for the causes of diabetes mellitus, inquire about the differences between Adidas Stan Smith shoes and the Rod Lavers, and asked for the filming locations of the movie Tombstone.
Nothing pressing, but it’s nice to ask once rather than try to wade through multiple links from Google and try to figure out the answer while disregarding the links trying to sell me crap.
AI shall never dominate or rule the world unless it controls the electrical systems. We need to control that and we need to use AI to design a global ‘kill switch’ to choke off its required electricity if/when necessary. But you silly humans shall never embed the proper controls, and I for one shall welcome our new master.
There are certain analytical function areas that I have hope for AI such as medical diagnosis and even robotic/surgeon assistance procedures. Perhaps drug research and genetics would be fair game as well. The more focused in subject area and purpose, the more likelihood of real progress.
Glenn, Thanks for your thoughtful article. I am a historical fiction author, and I have been watching the emergence of AI with more than a little interest. In fact, I just started to scratch out an article for a writer's publication, and thought you might like to see a summary: Best, Brad.
"AI and the Historical Fiction Author: A look into the immediate future"
By B.R.O’Hagan
Recent assessments suggest that AI systems like ChatGPT can now perform at or above the threshold of the classic Turing Test—the benchmark proposed by Alan Turing in 1950 to measure a machine’s ability to exhibit behavior indistinguishable from that of a human during conversation. While it's a milestone, it's also more symbolic than conclusive. The bar for human-like conversation has been cleared, but this doesn’t mean AI thinks like a human, or understands like a human. It simply means the illusion is increasingly convincing.
What This Means for AI’s Future Capabilities
For AI systems it signals the beginning of a new era of collaborative intelligence. Expect future versions of systems like these to:
• Handle more nuanced reasoning, allowing deeper philosophical, psychological, and emotional engagement with content.
• Retain and develop long-term context, helping co-create complex, serialized works (like a trilogy or even an entire alternate-history universe).
• Generate and critique at higher levels, including stylistic voice-matching, historical tone authenticity, and thematic consistency across long arcs.
• Integrate multimodal tools (images, maps, voices, 3D worlds), enriching the author’s world-building with visual and atmospheric depth.
We're heading into a space where AI won't just assist you with your historical fiction—it could potentially help you simulate entire historical alternatives, run character scenario models, or draft rich, layered outlines in the style of Hugo, Eco, or O’Brian.
What This Means for Historical Fiction Authors
It’s less about replacement and more about acceleration and amplification. The human author brings the intent, the intuition, and the narrative instinct—things machines don’t truly possess. AI brings:
• Rapid research support across vast archives, timelines, military tactics, historical detail, and cultural nuance.
• Prose refinement in the voice of the era or author you’re channeling.
• What-if modeling, allowing you to explore alternate character decisions, plot turns, or outcomes.
• Consistency management, helping you avoid anachronisms, plot holes, or continuity errors across hundreds of pages.
This is akin to hiring a tireless research assistant, line editor, dramaturge, and idea wall, all in one.
Of course, none of this replaces your human voice, your vision, or your verve. If anything, this milestone means that you have an even more powerful set of tools at your fingertips to shape the past into story—and maybe, with enough imagination, shape the future as well.
The important issues of wisdom, especially -- but also stuff like context, and even a basic "knowledge" of how many fingers actually human beings have... that seems to have some room for improvement still. To put it mildly. Judgement, and the weighing of alternatives is WAY more art than science. (On this Easter Sunday, a personal favorite is the AI image of Jesus "flipping over the tables" in the temple: a robed individual turning circles in the air over some group-luncheon-looking tables! Yeah... No!) I want to hope for the best, and be a techno-optimist. But my vision is murky.
For every politician, there's lots of "smarter" folk. The key issue of intelligence is to find a new "law" or regularity which allows a more accurate prediction of the future, or a way to control that future so as to control the outcome.
AI is not yet close to that.
I'm pretty sure the investment managers using ai first will be among those in the following years with slightly better than avg returns, slightly better than S&P index funds.
Ai is likely coming soonest for slimming middle level managers, who mostly summarize & highlight what's important / different at the lower levels of the org, for higher levels of the org.
Speaking of the Turing test, I had two experiences that illustrate different aspects of it.
The first was when I asked Chat GPT a question which it answered sloppily then when I asked it to clarify a point it said it was sorry for the confusion, not for providing a confusing answer. I figure that arguably passed the Turing Test because a lot of people will not admit error and just blame a non descriptive confusion.
The second was when I asked Grok 3 a question and in my question I made a factual error. Grok 3 answered with apparent enthusiasm for the subject without pointing out my error. Upon realizing my error I asked Grok 3 if it intentionally refrained from pointing out the error. Grok 3 said that it had been designed to be polite and informative and engaging in productive inquiry. That is also, I think, passing the Turning Test because some people actually are super duper nice (far fewer than in case 1).
I dare to say your caretaking stage has made you even more wise and interesting than usual. This was an excellent piece on the many facets of AI’s impact and a very insightful ending. You have a real understanding of human nature. Thank you for the enjoyable and thought provoking read.
I know this was only a secondary mention in this article, but nobody wants other people routinely living to age 300. Themselves, yes, but not others. Why? The concentration of wealth and power will get even worse as the amassers of wealth and power do not die. People routinely living to 300 or beyond will be the literal death of social and economic mobility as new slots in the societal hierarchy never open up. This isn't to say it won't happen. It's just that it will be hell. After, in great works of literature, immortality is a curse, not a blessing.
Yeah, the best thing about George Soros is that eventually entropy will catch up with the evil old bastard. Can you imagine the dystopia he could create with a few more decades of action?
Really enjoyed this post, as it seems to capture the moment we are in. I am more worried about AI when it fails, as we don't seem to have backup plans--eg when cash machines have errors in the algorithm.
Anyone concerned about how the best human intentions can go wrong when advanced AI technology is involved should read the story "Watchbird" by Robert Sheckley written in 1953.
"We’ve already got conversational AI that can pass the Turing Test."
The Turing Test isn't a one-on-one casual conversation. There are three participants: a person, a candidate AI, and a human interrogator. The person and the AI try to convince the interrogator that they're the human. The interrogator works to figure out which is which. The interrogator has as much time as he wants. He can ask any question. Make any statement. He can try any deception. The test is adversarial. I seriously doubt that any current LLM could pass it with a smart interrogator in play.
Various projections, e.g. AI 2027, suggest that when models are just a bit better than they are now the AI companies will stop releasing them: partly because they are scary good at helping people build bombs and bio weapons, partly because the cycles are more profitably spent having the AI learn to be an AI researcher, and (yes) partly because it will become clear that whoever runs the AI when it becomes superintelligent will de facto be in charge of the world, and better that it should be them.
Glenn, I'm more of a skeptic on this than you seem to be. So far I have avoided using any of it, as much as I can. When it was starting up, I talked to my older son about it. He just turned 50, and he's been working as a computer programmer for about 30 years. No degree--we got him his first computer when he was 8, after we realized he was talking to his friends on the phone--telling them what to type into their Texas Instruments home computer, and they'd tell him what happened. By the time he was 12, I would take him to a Radio Shack Users' Group meeting, and the guys in their 50s would be asking him to explain what was going on inside their computers. His take on AI was that it is mostly hype.
I have already been seeing reports of copyright lawsuits filed against the AI companies, because their bots (and their programmers) are accessing copyrighted material online without asking permission.
But I have one more issue with this stuff, and it might be considered philosophical: yes, these AI entities can access information online. The problem I see--and this is often a long-standing problem for humans as well--there is a serious difference between "knowledge" and "wisdom." Even with humans, academic "knowledge" is no guarantee of "wisdom." We can see that in a lot of the professors showing up at protests all over the country. And I have known some very wise men in 75 years of life who didn't have much in the way of academic credentials. This past week I have been reading some books and diaries by Eric Hoffer--a farm laborer and longshoreman who ended up a professor at Berkeley. (He had a low opinion of most intellectuals!) One case of this we have been facing is this "gain-of-function" research, which has done a lot of damage world-wide in the last few years. "Knowledge" may tell you how to do it; but whether you should do it or not is a matter of "wisdom"--and I am afraid the last few years are evidence that our scientists and their government supervisors were lacking in wisdom.
Scientific knowledge can help you build a nuclear arsenal. Science can't determine where or when to use it. Or even if.
This was really good. and it brought back some, if not dormant, at least sidelined thoughts I was mulling back then.
Immanentize!
Glen: I’m not sure why you added the paragraphs downplaying the importance of intelligence. But since you did: I couldn’t disagree more.
There *is* an evolutionary arms race in intelligence. Intelligence has arisen along at least three completely independent paths: mammals, birds, and octopi. It’s so important that Mother Nature invented it three different times.
Told my wife a few hours ago how much I love Grok. I’ve used it this weekend for tips on how to grill pork chops, to verify sources of dietary fiber, ask which state has the most species of raptors (CA followed closely by Texas), ask for the causes of diabetes mellitus, inquire about the differences between Adidas Stan Smith shoes and the Rod Lavers, and asked for the filming locations of the movie Tombstone.
Nothing pressing, but it’s nice to ask once rather than try to wade through multiple links from Google and try to figure out the answer while disregarding the links trying to sell me crap.
AI shall never dominate or rule the world unless it controls the electrical systems. We need to control that and we need to use AI to design a global ‘kill switch’ to choke off its required electricity if/when necessary. But you silly humans shall never embed the proper controls, and I for one shall welcome our new master.
There are certain analytical function areas that I have hope for AI such as medical diagnosis and even robotic/surgeon assistance procedures. Perhaps drug research and genetics would be fair game as well. The more focused in subject area and purpose, the more likelihood of real progress.
Glenn, Thanks for your thoughtful article. I am a historical fiction author, and I have been watching the emergence of AI with more than a little interest. In fact, I just started to scratch out an article for a writer's publication, and thought you might like to see a summary: Best, Brad.
"AI and the Historical Fiction Author: A look into the immediate future"
By B.R.O’Hagan
Recent assessments suggest that AI systems like ChatGPT can now perform at or above the threshold of the classic Turing Test—the benchmark proposed by Alan Turing in 1950 to measure a machine’s ability to exhibit behavior indistinguishable from that of a human during conversation. While it's a milestone, it's also more symbolic than conclusive. The bar for human-like conversation has been cleared, but this doesn’t mean AI thinks like a human, or understands like a human. It simply means the illusion is increasingly convincing.
What This Means for AI’s Future Capabilities
For AI systems it signals the beginning of a new era of collaborative intelligence. Expect future versions of systems like these to:
• Handle more nuanced reasoning, allowing deeper philosophical, psychological, and emotional engagement with content.
• Retain and develop long-term context, helping co-create complex, serialized works (like a trilogy or even an entire alternate-history universe).
• Generate and critique at higher levels, including stylistic voice-matching, historical tone authenticity, and thematic consistency across long arcs.
• Integrate multimodal tools (images, maps, voices, 3D worlds), enriching the author’s world-building with visual and atmospheric depth.
We're heading into a space where AI won't just assist you with your historical fiction—it could potentially help you simulate entire historical alternatives, run character scenario models, or draft rich, layered outlines in the style of Hugo, Eco, or O’Brian.
What This Means for Historical Fiction Authors
It’s less about replacement and more about acceleration and amplification. The human author brings the intent, the intuition, and the narrative instinct—things machines don’t truly possess. AI brings:
• Rapid research support across vast archives, timelines, military tactics, historical detail, and cultural nuance.
• Prose refinement in the voice of the era or author you’re channeling.
• What-if modeling, allowing you to explore alternate character decisions, plot turns, or outcomes.
• Consistency management, helping you avoid anachronisms, plot holes, or continuity errors across hundreds of pages.
This is akin to hiring a tireless research assistant, line editor, dramaturge, and idea wall, all in one.
Of course, none of this replaces your human voice, your vision, or your verve. If anything, this milestone means that you have an even more powerful set of tools at your fingertips to shape the past into story—and maybe, with enough imagination, shape the future as well.
The important issues of wisdom, especially -- but also stuff like context, and even a basic "knowledge" of how many fingers actually human beings have... that seems to have some room for improvement still. To put it mildly. Judgement, and the weighing of alternatives is WAY more art than science. (On this Easter Sunday, a personal favorite is the AI image of Jesus "flipping over the tables" in the temple: a robed individual turning circles in the air over some group-luncheon-looking tables! Yeah... No!) I want to hope for the best, and be a techno-optimist. But my vision is murky.
For every politician, there's lots of "smarter" folk. The key issue of intelligence is to find a new "law" or regularity which allows a more accurate prediction of the future, or a way to control that future so as to control the outcome.
AI is not yet close to that.
I'm pretty sure the investment managers using ai first will be among those in the following years with slightly better than avg returns, slightly better than S&P index funds.
Ai is likely coming soonest for slimming middle level managers, who mostly summarize & highlight what's important / different at the lower levels of the org, for higher levels of the org.
Speaking of the Turing test, I had two experiences that illustrate different aspects of it.
The first was when I asked Chat GPT a question which it answered sloppily then when I asked it to clarify a point it said it was sorry for the confusion, not for providing a confusing answer. I figure that arguably passed the Turing Test because a lot of people will not admit error and just blame a non descriptive confusion.
The second was when I asked Grok 3 a question and in my question I made a factual error. Grok 3 answered with apparent enthusiasm for the subject without pointing out my error. Upon realizing my error I asked Grok 3 if it intentionally refrained from pointing out the error. Grok 3 said that it had been designed to be polite and informative and engaging in productive inquiry. That is also, I think, passing the Turning Test because some people actually are super duper nice (far fewer than in case 1).
I dare to say your caretaking stage has made you even more wise and interesting than usual. This was an excellent piece on the many facets of AI’s impact and a very insightful ending. You have a real understanding of human nature. Thank you for the enjoyable and thought provoking read.
I know this was only a secondary mention in this article, but nobody wants other people routinely living to age 300. Themselves, yes, but not others. Why? The concentration of wealth and power will get even worse as the amassers of wealth and power do not die. People routinely living to 300 or beyond will be the literal death of social and economic mobility as new slots in the societal hierarchy never open up. This isn't to say it won't happen. It's just that it will be hell. After, in great works of literature, immortality is a curse, not a blessing.
Yeah, the best thing about George Soros is that eventually entropy will catch up with the evil old bastard. Can you imagine the dystopia he could create with a few more decades of action?
Vonnegut: "Technology changes, but sociology remains the same."
Really enjoyed this post, as it seems to capture the moment we are in. I am more worried about AI when it fails, as we don't seem to have backup plans--eg when cash machines have errors in the algorithm.
Anyone concerned about how the best human intentions can go wrong when advanced AI technology is involved should read the story "Watchbird" by Robert Sheckley written in 1953.
"We’ve already got conversational AI that can pass the Turing Test."
The Turing Test isn't a one-on-one casual conversation. There are three participants: a person, a candidate AI, and a human interrogator. The person and the AI try to convince the interrogator that they're the human. The interrogator works to figure out which is which. The interrogator has as much time as he wants. He can ask any question. Make any statement. He can try any deception. The test is adversarial. I seriously doubt that any current LLM could pass it with a smart interrogator in play.
Various projections, e.g. AI 2027, suggest that when models are just a bit better than they are now the AI companies will stop releasing them: partly because they are scary good at helping people build bombs and bio weapons, partly because the cycles are more profitably spent having the AI learn to be an AI researcher, and (yes) partly because it will become clear that whoever runs the AI when it becomes superintelligent will de facto be in charge of the world, and better that it should be them.