An AI-generated podcast; Mustafa Suleyman on the latest; Tim B. Lee on a weakness in LLMs; and on another weakness; James Pethokoukis on the link to robots
"I want the LLM to differentiate what it “knows” from how it converses. I think it’s fine if it learns to talk to economists by seeing how they talk to one another. But I want it to forget the substance of the talk and instead to learn how I think. That isn’t how training works."
I will reiterate my proposition that how one arranges language patterns and how one is able to "think" are not nearly separable, and that only a small percent of humans are actually able to think creatively beyond mimicry of the patterns to which they have been exposed.
In my own experience, this "not forgetting the substance of the talk" is actually how a lot of human brains work, at the "smart-but-not-smart-enough" level of a conversation in which they want to participate. One sees this all the time in online forums. Maybe the "know enough to be dangerous" level. It is a kind "high-functioning hollow" state in which a person has been exposed to so much of the genuine article as a spectator that they have indeed learned enough of the surface-level ideas and patterns of linguistic combinations to almost pass a kind of "Intellectual Turing Test" passing for a genuinely intelligent person not intimately familiar with the subject matter down to its roots, but -able- with it, in the sense of being able to work things out at a high level.
But they haven't really. As per "those who can, do, those who can't, teach" one encounters many excellent high school teachers at this level but not higher. As soon as you try to probe for more detail or extend creatively into novel concepts they reveal themselves to lack the essential elements that would make them capable of contribution to the discussion or genuine "thought" about the issue at hand. They are doing the linguistic equivalent of "going through the motions", and like the members of a Cargo Cult, they think they are doing something meaningful and do not even really understand that they are just going through motions.
A long time ago at Less Wrong there were discussions about Feynman's "Cargo Cult Science" and also "Cargo Cult Programming" and "Cargo Cult Language" and other extensions of the concept of high-functioning imitations yet still devoid of the degree of conceptual understanding and facility that would, for example, allow a programmer copying form from real masters and dropping in libraries from being able to know enough to fix a subtle bug buried deep in the code's structure. (Arranging supervisory hierarchies and quality control in order to deal with this trade-off between scarce expertise and the need to delegate to Cargo-Cult subordinates, combined with Conway's Law, is one of my pet working theories on why large, old organizations rot).
My point is that the LLM's make -great- teachers but the level of a teacher that still can't "peel back" the curriculum (i.e., "conventional wisdom") enough to successfully build up a model of someone with a contrarian or heterodox worldview. Just like most humans can't.
I have a feeling those guesses on humanoid robot use by 2050 are not going to age well. I suppose 25 years is a long way off, though, so who knows. Still feels like another one of those technological dreams that is going to be 10 years away for the next century, however.
Pethokoukis parenthetically mentions a study about robots in Japanese nursing homes, a staple of robot journalism since about 2014 when it seems to have become official industrial policy. Curious, I asked an AI about this and it returned:
"Japan's industrial policy and the Ministry of Economy, Trade and Industry (METI) have played significant roles in the introduction of robots to nursing homes, addressing the country's demographic challenges of an aging population and labor shortages. Here's how:
Government Support and Strategy: METI has been instrumental in promoting robot technology in nursing care through various strategic plans and financial support. For instance, under initiatives like the "New Robot Strategy" and the "Five-year Plan for Developing Nursing Care Robots," Japan has aimed to become a leader in robotics, particularly in sectors like eldercare. These strategies have included the development and introduction of robots specifically designed for nursing care, aiming to improve quality of life for the elderly and alleviate the burden on human caregivers.
Subsidies and Funding: METI has facilitated the adoption of robots by providing financial support. Local governments also offer subsidies for adopting robots in nursing homes, with some prefectures subsidizing up to 50% of the cost per robot. This financial assistance has been crucial in making the high cost of robotics technology more manageable for care facilities. By 2018, many prefectures were actively supporting these initiatives, which has helped in the broader implementation of robotic solutions.
Regulatory Framework and Safety Standards: METI has worked on deregulating to ease the introduction of robots while ensuring safety. This includes revising certification procedures for medical devices and pushing for international safety standards for care robots, which has allowed for more practical and widespread use of robotics in nursing homes. For example, Panasonic's robotic nursing bed was one of the first to clear such standards, facilitating its use in care settings.
Impact on Employment and Care Quality: Studies, including those from Stanford University's Asia-Pacific Research Center, have indicated that robot adoption in Japan's nursing homes has not reduced employment as feared but rather has increased opportunities for non-regular care workers, contributing to staff retention and flexibility. This suggests that METI's policies have indirectly supported labor market dynamics in the care sector by integrating technology in a way that complements human work rather than replacing it.
Demonstration and Promotion: METI has also been involved in large-scale demonstrations to show the practical application of care robots, which has helped in promoting the technology both domestically and internationally. The ministry's efforts have led to Japan being viewed as a pioneer in eldercare robotics, influencing global markets and policy-making in similar demographic situations.
In conclusion, Japan's industrial policy through METI has been pivotal in shaping the landscape for robotics in nursing homes. By combining strategic planning, financial incentives, regulatory adjustments, and promotion, METI has significantly influenced how robots are integrated into eldercare, although the adoption rate and real-world impact still show room for growth and further study."
More generally, AI and robotics both seem to be creatures of industrial policy. Per the same AI:
Industrial policy has played a significant role in shaping the growth of artificial intelligence (AI) and robotics across various countries, influencing both technological development and economic outcomes. Here's how:
Investment and Strategic Focus: Many countries have implemented national strategies that prioritize AI and robotics as key areas for future economic and technological competitiveness. For example, China's "New Generation Artificial Intelligence Development Plan" aims to make China a world leader in AI by 2030, encouraging significant investments in AI research, development, and application across industries like manufacturing, healthcare, and transportation. This strategic focus has led to China becoming the world's largest market for industrial robots since 2013, which in turn has had a substantial impact on labor markets and industrial production efficiency.
Regulatory Environment and Incentives: Governments have used industrial policies to create favorable regulatory environments, including tax incentives, subsidies, and grants, to stimulate private sector investment in AI and robotics. For instance, the United States, through initiatives like the Advanced Manufacturing Leadership Strategy, supports the integration of AI into industrial processes, aiming to boost productivity and bring manufacturing back to developed countries. Such policies can encourage innovation but also pose challenges in terms of managing the displacement of labor and ensuring ethical AI development.
Education and Workforce Development: Industrial policies have also focused on education and skills development to prepare the workforce for an AI-driven economy. Programs aimed at retraining workers for jobs that complement AI systems or for roles in AI development itself are crucial. This aspect of policy aims to mitigate potential job displacement effects of automation while fostering a skilled labor market that can support and expand AI applications. For example, the emphasis on STEM education and the integration of liberal arts with technology education in some countries helps in developing the kind of intuitive thinkers needed for AI innovation.
Global Competitiveness and Economic Growth: By integrating AI and robotics into industrial policy, countries aim to enhance their global competitiveness. Policies that support AI can lead to higher productivity, new job creation in tech sectors, and the transformation of traditional industries. However, this also introduces risks of increasing inequality between countries, as those at the forefront of AI technology might attract more investment and skilled labor, potentially widening the economic divide. This dynamic is highlighted by research indicating that AI could drive economic divergence if not managed with inclusive policies."
No doubt this is uncharitable, but it sounds like part of the "industrial policy" favoring development and use of robots is "we won't smother them with regulations like we do so many other things."
I have not partaken in either producing or consuming videos or podcasts for many years now. Reading, for me at least, is just so much more efficient. HeyGen does not seem as if it will be enough to persuade me to rectify my failure. But, if it substantially reduces production costs perhaps it will launch a thousand new stars whose voices will eventually find their way into print and so that would be of personal benefit.
Well, Dr. Kling explicitly invited comment on his AI generated podcast so my comment on how AI is not going to cure my aversion seemed relevant to me, but apologies for offending you.
Don't take this the wrong way but I think you may have developed a reputation for prickliness, for seeming pissed off that other commenters have not been responsive to you and/or have moved the thread away from a direction you wanted it to go.
Watching and listening to the robots restores my appreciation for humanity. The intentional "uhs" and vacuous clichés are technical achievements but pitifully pathological.
The video was mildly interesting between two amenable conversationalists, but it would be more engaging if an actual argument, with emotion, developed from differing points of view. Perhaps a paper written that way would generate such an AI encounter. Maybe there's examples of this somewhere.
"I want the LLM to differentiate what it “knows” from how it converses. I think it’s fine if it learns to talk to economists by seeing how they talk to one another. But I want it to forget the substance of the talk and instead to learn how I think. That isn’t how training works."
I will reiterate my proposition that how one arranges language patterns and how one is able to "think" are not nearly separable, and that only a small percent of humans are actually able to think creatively beyond mimicry of the patterns to which they have been exposed.
In my own experience, this "not forgetting the substance of the talk" is actually how a lot of human brains work, at the "smart-but-not-smart-enough" level of a conversation in which they want to participate. One sees this all the time in online forums. Maybe the "know enough to be dangerous" level. It is a kind "high-functioning hollow" state in which a person has been exposed to so much of the genuine article as a spectator that they have indeed learned enough of the surface-level ideas and patterns of linguistic combinations to almost pass a kind of "Intellectual Turing Test" passing for a genuinely intelligent person not intimately familiar with the subject matter down to its roots, but -able- with it, in the sense of being able to work things out at a high level.
But they haven't really. As per "those who can, do, those who can't, teach" one encounters many excellent high school teachers at this level but not higher. As soon as you try to probe for more detail or extend creatively into novel concepts they reveal themselves to lack the essential elements that would make them capable of contribution to the discussion or genuine "thought" about the issue at hand. They are doing the linguistic equivalent of "going through the motions", and like the members of a Cargo Cult, they think they are doing something meaningful and do not even really understand that they are just going through motions.
A long time ago at Less Wrong there were discussions about Feynman's "Cargo Cult Science" and also "Cargo Cult Programming" and "Cargo Cult Language" and other extensions of the concept of high-functioning imitations yet still devoid of the degree of conceptual understanding and facility that would, for example, allow a programmer copying form from real masters and dropping in libraries from being able to know enough to fix a subtle bug buried deep in the code's structure. (Arranging supervisory hierarchies and quality control in order to deal with this trade-off between scarce expertise and the need to delegate to Cargo-Cult subordinates, combined with Conway's Law, is one of my pet working theories on why large, old organizations rot).
My point is that the LLM's make -great- teachers but the level of a teacher that still can't "peel back" the curriculum (i.e., "conventional wisdom") enough to successfully build up a model of someone with a contrarian or heterodox worldview. Just like most humans can't.
I have a feeling those guesses on humanoid robot use by 2050 are not going to age well. I suppose 25 years is a long way off, though, so who knows. Still feels like another one of those technological dreams that is going to be 10 years away for the next century, however.
I find the video congenial, but there is an irony. Neither interlocutor tries to steel-man a case for adversarial argument.
So, just like any normal podcast between humans.
Pethokoukis parenthetically mentions a study about robots in Japanese nursing homes, a staple of robot journalism since about 2014 when it seems to have become official industrial policy. Curious, I asked an AI about this and it returned:
"Japan's industrial policy and the Ministry of Economy, Trade and Industry (METI) have played significant roles in the introduction of robots to nursing homes, addressing the country's demographic challenges of an aging population and labor shortages. Here's how:
Government Support and Strategy: METI has been instrumental in promoting robot technology in nursing care through various strategic plans and financial support. For instance, under initiatives like the "New Robot Strategy" and the "Five-year Plan for Developing Nursing Care Robots," Japan has aimed to become a leader in robotics, particularly in sectors like eldercare. These strategies have included the development and introduction of robots specifically designed for nursing care, aiming to improve quality of life for the elderly and alleviate the burden on human caregivers.
Subsidies and Funding: METI has facilitated the adoption of robots by providing financial support. Local governments also offer subsidies for adopting robots in nursing homes, with some prefectures subsidizing up to 50% of the cost per robot. This financial assistance has been crucial in making the high cost of robotics technology more manageable for care facilities. By 2018, many prefectures were actively supporting these initiatives, which has helped in the broader implementation of robotic solutions.
Regulatory Framework and Safety Standards: METI has worked on deregulating to ease the introduction of robots while ensuring safety. This includes revising certification procedures for medical devices and pushing for international safety standards for care robots, which has allowed for more practical and widespread use of robotics in nursing homes. For example, Panasonic's robotic nursing bed was one of the first to clear such standards, facilitating its use in care settings.
Impact on Employment and Care Quality: Studies, including those from Stanford University's Asia-Pacific Research Center, have indicated that robot adoption in Japan's nursing homes has not reduced employment as feared but rather has increased opportunities for non-regular care workers, contributing to staff retention and flexibility. This suggests that METI's policies have indirectly supported labor market dynamics in the care sector by integrating technology in a way that complements human work rather than replacing it.
Demonstration and Promotion: METI has also been involved in large-scale demonstrations to show the practical application of care robots, which has helped in promoting the technology both domestically and internationally. The ministry's efforts have led to Japan being viewed as a pioneer in eldercare robotics, influencing global markets and policy-making in similar demographic situations.
In conclusion, Japan's industrial policy through METI has been pivotal in shaping the landscape for robotics in nursing homes. By combining strategic planning, financial incentives, regulatory adjustments, and promotion, METI has significantly influenced how robots are integrated into eldercare, although the adoption rate and real-world impact still show room for growth and further study."
More generally, AI and robotics both seem to be creatures of industrial policy. Per the same AI:
"
Don J <djjinva79@gmail.com>
10:13 AM (21 minutes ago)
to me
Industrial policy has played a significant role in shaping the growth of artificial intelligence (AI) and robotics across various countries, influencing both technological development and economic outcomes. Here's how:
Investment and Strategic Focus: Many countries have implemented national strategies that prioritize AI and robotics as key areas for future economic and technological competitiveness. For example, China's "New Generation Artificial Intelligence Development Plan" aims to make China a world leader in AI by 2030, encouraging significant investments in AI research, development, and application across industries like manufacturing, healthcare, and transportation. This strategic focus has led to China becoming the world's largest market for industrial robots since 2013, which in turn has had a substantial impact on labor markets and industrial production efficiency.
Regulatory Environment and Incentives: Governments have used industrial policies to create favorable regulatory environments, including tax incentives, subsidies, and grants, to stimulate private sector investment in AI and robotics. For instance, the United States, through initiatives like the Advanced Manufacturing Leadership Strategy, supports the integration of AI into industrial processes, aiming to boost productivity and bring manufacturing back to developed countries. Such policies can encourage innovation but also pose challenges in terms of managing the displacement of labor and ensuring ethical AI development.
Education and Workforce Development: Industrial policies have also focused on education and skills development to prepare the workforce for an AI-driven economy. Programs aimed at retraining workers for jobs that complement AI systems or for roles in AI development itself are crucial. This aspect of policy aims to mitigate potential job displacement effects of automation while fostering a skilled labor market that can support and expand AI applications. For example, the emphasis on STEM education and the integration of liberal arts with technology education in some countries helps in developing the kind of intuitive thinkers needed for AI innovation.
Global Competitiveness and Economic Growth: By integrating AI and robotics into industrial policy, countries aim to enhance their global competitiveness. Policies that support AI can lead to higher productivity, new job creation in tech sectors, and the transformation of traditional industries. However, this also introduces risks of increasing inequality between countries, as those at the forefront of AI technology might attract more investment and skilled labor, potentially widening the economic divide. This dynamic is highlighted by research indicating that AI could drive economic divergence if not managed with inclusive policies."
Maybe food for thought.
No doubt this is uncharitable, but it sounds like part of the "industrial policy" favoring development and use of robots is "we won't smother them with regulations like we do so many other things."
I have not partaken in either producing or consuming videos or podcasts for many years now. Reading, for me at least, is just so much more efficient. HeyGen does not seem as if it will be enough to persuade me to rectify my failure. But, if it substantially reduces production costs perhaps it will launch a thousand new stars whose voices will eventually find their way into print and so that would be of personal benefit.
I'm not sure what your podcast aversion has to do with AI.
Well, Dr. Kling explicitly invited comment on his AI generated podcast so my comment on how AI is not going to cure my aversion seemed relevant to me, but apologies for offending you.
I have no idea why you think my uncertainty implies offense taken.
Don't take this the wrong way but I think you may have developed a reputation for prickliness, for seeming pissed off that other commenters have not been responsive to you and/or have moved the thread away from a direction you wanted it to go.
Yes, thanks. All of that may be true but it doesn't change anything I wrote on this thread.
Watching and listening to the robots restores my appreciation for humanity. The intentional "uhs" and vacuous clichés are technical achievements but pitifully pathological.
I agree with GPT4o's review of my review: https://chatgpt.com/share/6769767a-58e4-8008-b48f-f6ca81d87fe1
The video was mildly interesting between two amenable conversationalists, but it would be more engaging if an actual argument, with emotion, developed from differing points of view. Perhaps a paper written that way would generate such an AI encounter. Maybe there's examples of this somewhere.