The 69 Controversies of AI Adoption

Toyota: Supra Intelligence Some Shoppers: SuperStop and Go Intelligence Phonies: Superficial Intelligence Government Workers: SuperOfficial Intelligence Military: SuperCommissioned Officer Intelligence Governments: Superpower Intelligence Children: Superhero Intelligence

Today I read a lot of negative reaction to this post by an Anthropic employee about NVidia. However, this is actually a rational and expected argument from an economic or strategic perspective as explained in The Last AI @The_Last_AI and the 69 Controversies of AI Adoption @69AIControversy and here's why: 1) Pespective 1: US vs China in the race towards AGI. If we consider getting to AGI to be something meaningful, then it would hinder companies in free market economies develop it if the profitability of the industry went down. Planned economies such as China can induce this lower profitability of the industry easily by giving directions and subsidies to its companies to give products away for free, or open-source. Some may call it dumping, or selling below price. This explains why we would expect to see more open-weights from China. It helps collapse the profitability model of US firms that live in the market ecosystem. Slows US down, accelerates China. 2) Perspective 2: Software vs Hardware Companies. Within US, we may also think of AI developers and hardware manufacturers as suppliers of each other, such that each prefer the price of the other to be lower for own profitabilty. This explains why there is incentive for AI developers such as OpenAI @sama Google @sundarpichai and Anthropic @DarioAmodei to also invest in hardware to lower their cost, in addition to gaining other advantages in the race towards AGI. We see Elon Musk is amicable towards open-source, because his firms are more hardware-oriented. In this perspective, it is perpectly expectable to see two opposing preferences, as demonstrated by Jensen Huang to prefer open-source of AI models, while Anthropic employee to prefer open-source of hardware-related proprietary knowledge owned by Nvidia such as CUDA. What people who are criticizing this post are missing may be this layer of the competitive forces that affect these industries. Looking forward, what may additionally complicate this matter would be how China has advantages in physical AI. We may even see open-source versions of CUDA competitor if Chinese manufacturers reach a certain level of competitiveness against Nvidia. This is probably what may keep @JensenHuang busy now more than any other competitive force.

Yes, and I hope the applicants would have read "The Last AI," the book that introduces AI Economics!

This is an interesting proposal not only as an alternative way to raise funding, but to possibly address a humanity-scale catastrophe that Elon Musk likes to emphasize. 1. On Elon Musk To summarize the problem outlined in The Last AI (@The_Last_AI) and the 69 Controversies of AI Adoption (@69AIControversy), we may ask the following question: What if Earth for some reason becomes suddenly uninhabitable and moving go Mars is the only option left. Who should be chosen to go to Mars and survive? Available seats would be highly limited. This may be similar to the Titanic problem, where women and children were favored, or the Noah's Arc problem, where one couple of each species was spared. In modern societies, if this problem is to be solved by private corporations (as opposed to by the government), then the solution would be to choose from the stockholders of the company. If what some say about SpaceX, that it does not need more funding or need to go public, is true, then we may speculate an alternative reason for Elon Musk to go public is to give this opportunity or perk to a wider base of people to go to Mars in case of some emergency. Since Tesla shareholders would be more likely to share Elon's vision, they may be better candidates to purchase SpaceX shares. In short, if one believe the missiom of SpaceX is to save humanity from some unforeseen danger, then having a larger ownership base may mean something. 2. On Bill Ackerman Here are a few takes on potential questions to this post: - Why this instead of IPO? While IPOs are the standard way to go public, it costs more. The initial buyers of shares may not be as widespread as the proposed method. - What's in it for Pershing Square? It does gain some business because it would probably not be selected to lead arrange the IPO anyway. - Will current SpaceX shareholders lose or gain? This depends on the eventual valuation. In the sense that this method costs less, it may be favorable. Having more publicity may help raise the valuation. All in all, Ackerman's proposal may give more publicity to SpaceX and Tesla than regular IPO. Add in the potential reason for going public as saving humanity and Elon's distaste for IPO, Ackerman's proposal may offer an attractive alternative. --- On a side note, for anyone interested in how AI may change our business, economy, amd society, check out "The Last AI!"

1) For anyone interested in how different people may think differently about AI, check out the 69 Controversies of AI Adoption. amazon.com/dp/B0F9WZWPWB/ 2) For anyone interested in describing the transition between augmentation and automation, check out this paper on AI Effortlessness: sciencedirect.com/science/ar… 3) For anyone interested in how AI adoption may progress in our business, economy, and society,check out this paper on the Three Levels of AI Adoption and the Six Levels of Autonomous Companes, papers.ssrn.com/sol3/papers.… 4) and The Last AI amazon.com/dp/B0D5B63LSV/

Trying to use AGI, which is a technology-centered concept, as a measure for AI adoption, which is more business-oriented concept, in itself causes confusion. This is the reason for introducing the concepts of the "3 Levels of AI Adoption" and "AI effortlessness. " The 3 Levels of AI Adoption: The important milestone equivalent to AGI is "AI Equalization," to reach Level 2 AI Adoption, which is when AI reaches parity for each job humans were carrying out. This way, whereas AGI will be defined by AI "suppliers," Lv2 AI Adoption will be defined by AI "demand," or the users of AI. Lv3 AI Adoption is when AI takes charge of the whole firms. This concept later became famous by OpenAI's 5 Stages of AI. Anyone wondering about Stages 2 and 4 (AI that innovates) may read about the "enablers" and "the 6 Levels of Autonomous Companies" in the same paper. AI Effortlessness We have categorized AI Adoption as either augmentation or automation. However, we may also place both types of AI adoption into the same dimension by measuring how AI reduces human effort. In augmentation, human effort is not reduced completely, while automation can completely eliminate effort. In the dimension of AI Effortlessness, augmentation is expressed as a number between 0 and 1, and automation (replacement automation) is assigned as 1. This also allows applying of the concepts of "Automated Innovation." In "Advance Automation," where AI Effortlessness is above 1, AI with innovative capabilities would improve on the tasks it was previously carrying out, achieving "innovation" by AI itself. Whereas people may think automation is just replacing humans, we may think of it as evetually "advancing" the work in some sense. The paper also provides few examples of how these progresses may be meaningful for firms and the economy. I am excited to hear about this project and hope some of these concepts from my work could be incorporated in the new measure! Due to X policy, I'll post links in the reply.

One direction is to emphasize how people may have opposite opinions about AI, as explained in The 69 Controversies of AI Adoption. Link in the reply due to X policy

Yes, and the answers are already available, and based on the interviews, Sam Altman appears to have already read "The Last AI," so OpenAI probably knows it as well. As explained in my article, this 5 Stages of AI is a way to explain to internal stakeholders the "3 Levels of AI Adoption and the 6 Levels of Autonomous Companies." The Stages 1, 3, 5 correspond to AI Adoption Levels 1,2,3 and Stages 2,4 are "enablers." (One of the reasons some people are confused about Stage 4) What happens once these levels are achieved is explained in my book "The Last AI" which also introduces the 4 Levels of AI Adoption for society. Once you read it, you will understand why only 4 levels are necessary to get to the last AI. Links on my profile (due to X policy on links)

After reading recent posts about Mathematicians vs Physicists, Physicists vs Economists, Economics vs Finance, etc., here's an observation about Economics vs Business Management on AI Adoption. For those who do not know me, I majored both in Economics and Business so hopefully I am not too biased. 1) Economists have to make assumptions to simplify their analyses. Otherwise, things would become too complicated to model. Because of this, it appears economists tend to be more likely to just pick one side they like and focus more on that side, either on augmentation or automation. For example, an economist can assume AI will advance little, and focus on using AI as a tool, or advance a lot and call it "transformational AI" and begin their anaysis from there. 2) Business management researchers need to go one step further down and give more thoughts to understanding what would advance AI and how it may affect how businesses make decisions inside the firm. An example of this is the Three Levels of AI Adoption for management and the Four Levels of AI Adoption for society. In this framework you need to consider both augmentation and automation and how and why they may transition from one to the other and what may cause friction or resistance to switch. In short: if Economists find it easier to focus on either augmentation or automation, Management researchers may find it easier to focus on the transition between augmentation and automation. For More: Differences of perspectives similar to this are abound around us. For anyone interested in how different people may think differently about AI, check out The 69 Controversies of AI adoption. (Links on the next post)