Human → Superhuman → Ultrahuman

🌈 Abstract

The article discusses the evolution of AI's capabilities in chess, from being initially inferior to humans, to surpassing human grandmasters, and eventually reaching a point where humans can no longer contribute to the game. It explores the different phases of AI development - subhuman, human, superhuman, and ultrahuman - and how this pattern applies not just to chess, but to various domains as technology advances. The article also touches on the philosophical implications of AI's superiority in certain tasks, and the need to redefine our understanding of human capabilities and identity.

🙋 Q&A

[01] The Superhuman Phase in Chess

1. What was the first sign that computers were no longer just "human level" at chess? The first sign was Garry Kasparov's famous loss against Deep Blue in 1997, which prompted him to invent Advanced Chess, where humans and computers teamed up as equals.

2. When did the superhuman phase in chess officially begin? The superhuman phase in chess officially began in 2005, when the last time a grandmaster dared challenge an AI and was obliterated.

3. What happened as Stockfish and AlphaZero became too good for humans to contribute to tactics or strategy? As Stockfish and AlphaZero became too good, wins were AI's wins, and losses were often downstream of human mistakes. Even praised grandmasters, including Magnus Carlsen, became a hindrance for computers.

4. What was the likeliest outcome by 2020 in chess matches between computers and human-computer cyborgs? By 2020, "draw-death" was the likeliest outcome, as the human contributions were zero.

5. How long did the superhuman phase in chess last? The superhuman phase in chess began in 1997 and ended in 2020, lasting a little more than 20 years.

[02] The Phases of AI Development

1. What are the four phases of AI development described in the article? The four phases of AI development described are: subhuman, human, superhuman, and ultrahuman.

2. How does the article describe the pattern of AI's advancement in different domains? The article states that as automation and AI advance in any field, it will first find a task impossible, then gradually become capable of doing it at all, then eventually capable of better than many or most humans, and then better than the best human. This implies a further level of skill, where no human is able to improve the AI's results at all rather than get in the way or harm it.

3. What is the author's view on the "interstitial phase" during which humans and AIs collaborate as equals? The article suggests that the interstitial phase during which humans and AIs collaborate as equals cannot be eternal, as AI, computing, and robotics keep challenging the corners of our supremacy, bringing down the pillars of our identity as top-of-the-chain.

4. What is Moravec's paradox, and how does it relate to the different phases of AI development? Moravec's paradox suggests that those tasks that seem more robust, because they take more effort on our part, are the first to fall to AI. The article uses this to explain that chess feels hard but is easy, while walking feels easy but is hard.

[03] The Implications of Ultrahuman AI

1. What is the author's view on the implications of AI becoming "ultrahuman" in various domains? The author suggests that as technology conquers more and more domains, the end picture unfolds as "ultrahumanity," where AI surpasses human capabilities in various tasks. The author questions whether there is anything that won't be conquered by ultrahuman AI.

2. How does the author suggest we should view the relationship between humans and ultrahuman AI? The author suggests that we should stop thinking in terms of optimization, improvement, and being the best, and instead focus on the dimensions of humanity that we care about, such as love, sympathy, and humanity itself. The author argues that granting the label "superhuman" to machines that lack the most vital dimensions of humanity can obscure the very things about being human that we care about.

3. What is the author's view on the importance of human imperfections and limitations? The author suggests that it's time to stop thinking of ourselves as kings losing our kingdoms and instead embrace and relish the imperfections that made us. The author argues that sometimes, what we want is an equally imperfect human, and that humans like humans.