Having been an avid podcast listener the last few years and always an eager reader, I now have a feeling that many topics, debates, and conclusions are being recycled over and over again. Also the mix of podcast topics (science, politics, world affairs, philosophy) and selection of podcast guests have started to converge beween the podcasts (I am probably also being lead to the same type of podcasts by an algorithm which accentuates the problem). Often A is a guest in B’s podcast and the reciprocates by having B as a guest in their podcast. Discussing many of the same topics. The Trump presidency didn’t help either as it is a gift to podcasters that keeps on giving.
Regardless of the mechanism, I now feel that many topics are overdiscussed. I don’t think that science or moral inquiry have stopped but some topics seem have hit a cul de sac and there is not much new to say—for now.
I admit that I might have grown too strong priors about some topics as well and that I’m therefore unwilling to take in new evidence even if it there. Be it as it may, these are some of the topics that I’m tired of now:
Free will: As a naturalist I think this debate should have been settled long ago. Free will as a concept cannot even be defined. Free from what? At a macro level there seems to be an arrow of time and an arrow of entropy in the universe. There is cause and effect. There is determinism and there is perhaps randomness. The Buddhists were right over 2000 years ago when they posited that everything (except Nirvana) is conditioned.
No-self: The self is as impossible to define as free will. The self is a useful illusion that keeps us alive, but nevertheless an illusion. The same Buddhists that rejected unconditioned volition also wisely rejected the idea of the self. They used the term anattā for no-self. (While I think the topic is pretty much settled, I am now reading Thomas Metzinger’s book The Ego Tunnel as I’m curious about his thinking in general.)
AI consciousness: There is a host of well-known scholars that seem to say that LLMs have become conscious. The claim usually doesn’t come with a definition of consciousness or a method to establish the presence of consciousness. LLMs may exhibit echoes of self-awareness but it is as reliable as their other responses. The as of yet unexplained kind of consciousness is the ability to have subjective experiences, to feel pleasure and pain. And indeed feel separate from the rest of the universe. Science seems stuck regarding this “hard problem of consciousness”. Pleasure and pain and the self have utility and are most likely the products of top-down evolutionary pressure. They may very well require a (partly) analog substrate like the brain (see this post). I’d put the probability for consciousness (in the form of subjective experiences) to emerge out of NAND-gates on par with the discovery of telepathy, lower than that of encountering aliens.
Religion: In Iceland, road construction projects are sometimes delayed, altered, or rerouted to avoid disturbing rocks believed to be the homes of huldufólk (hidden people or elves). Most people probably find this quaint but not awfully consequential. To me, all theological debates are about huldufólk. Maybe occasionally witty or quaint but of no consequence. There is a discussion to be had about religious extremism but that is a meta-level discussion that doesn’t concern the ontological status of divine beings per se. It’s a debate about harmful ideas, whatever those ideas might be: islamism, fascism, racism, …
Emergence: This seems to be the go-to explanation for all kinds of phenomena when other explanations fail. It is a frequently misused concept. My prior is that all phenomena must have a physical manifestation. A substrate. Consciousness, mathematics, human behavior, moral truths, and everything else need a substrate to exist. A Platonic “realm of forms” would require the discovery of an entirely new kind of physics. Also, low-entropy structures very seldom “emerge” at random from substrate even if it may happen if you give it a few billion years and a big universe. And even so, we seldom see excavators spontaneously emerging from iron ore or Bolzmann brains popping out of thin air. Substrate needs external influence or an improbable coincidence to organize itself. See this post.
AIG: I saw this headline the other day: ‘Sam Altmant is excited for AI to “solve all of physics”‘. We today don’t know what 95% of the universe consists of. It is highly unlikely that an AI will figure that out sitting in a data center, only based on the knowledge it has been given in its training data. This would require that all human knowledge is already implict in that training data. The only way we know of today to create new knowledge is by interacting with the universe using the scientific method. See this post and this post.
AI-risks: Humans are orders of magnitude more dangerous than AI models. Evil humans don’t need AI models to do evil deeds. A nuclear weapon is quite enough. AI intelligence is also not the same as AI will. Malicious intentions do not arise from intelligence but from poor engineering. AI application risks can therefore be managed with good engineering practices and regulation depending on the application. Many areas such as aviation, nuclear energy, medical devices, food and drugs are already regulated for our safety. See this post.
But I can be wrong. Please let me know!