16 ms·
> You say that "the community" derives facts or claims from unproved assumptions, yet at the same time you say that you "strongly tend to disagree" with those t
by f_klem 3mo ago
> You say that "the community" derives facts or claims from unproved assumptions, yet at the same time you say that you "strongly tend to disagree" with those theories and that the theories are "flawed in the sense that they cannot account for subjective experience and agency, amongst other things" merely on account that they are neither confirmed nor unconfirmed. I am confused about your stance. You allow yourself to have strong opinions about something unknown yet criticize other people for the same.
The assumptions I refer to are not only unproved, there is also increasing evidence that they are false. I do not criticize based on the assumption that there is subjective experience, but on the well developed idea that there must be something like 'subjective experience'. Here we enter the realm of philosophy, which by the way, is what science encounters when it runs out of answers. And this was precisely my point: AI research is based on assumptions that _need support or help_ from philosophy, not only neuroscience. But what is at stake here is the prevailing neurocentrism and scientificism characteristic of our era.
> I think it is absolutely normal that the core of a theory is based on not directly testable assumptions. And it's normal that people push it forward if it bears fruits, that's not a fallacy in any way, that's normal inquiry that may or may not lead to successful results.
That is correct and it is precisely why they are called 'theories': because the evidence points towards a specific direction but there is not yet enough evidence to call it a law.
Yet different theories, based on different assumptions, demand that those assumptions be tested at the fundamental level: logical, epistemological, philosophical, etc.
Regarding the theory that current LLM research could lead to human level intelligence, many people have the opinion that it can be discarded on fundamental grounds. Why? Because the assumptions that this theory stands on are flawed.
An issue I repeatedly see in the community (about which Dreyfus already wrote in his 1972 book, confirmed in his 1992 book, and we still see today) is that challenging the fundamental flaws in which current AI research is based on immediately sparks outrage in the AI community, as if people challenging those assumptions are against AI or AI research at all. I think that is a really silly, childish and not very humble position, and ultimately slows down research.
- LudwigNagasena 3mo agoNow you again say that there is increasing evidence that the assumptions are wrong and that the foundation is flawed, but when you get to specifics you merely claim that something is unknown or unconfirmed.
- f_klem 3mo agoThe books that I referenced at my first comment already contain an extensive explanation of why those assumptions are flawed, false, or ungrounded. I will not paste here parts of books. Nonetheless, I've been compiling references on different AI research assumptions and problems. I'll paste them here later on.
- LudwigNagasena 3mo agoIf I take a text written on a tangential topic from a generation or two ago and try to imagine how it applies to the current state of AI that would be me putting words in your mouth and speculating on your interpretation. I don't want to waste time going over weak uncompelling critique like, if human ~~sound production~~ intelligence is analogue, then ~~digital speakers~~ artificial intelligence is unlikely to be possible (to paraphrase the critique of the Biological Assumption); but I am genuinely interested in the increasing evidence that shows that the foundation of modern AI research is flawed.
- ozgung 3mo agoAfter watching a lecture of Dreyfus on Heidegger and skimming his books I think I begin to see what's going on. Hubert Dreyfus teaches a specific set of philosophers beginning with Heidegger. His brother Stuart is an industrial engineer working on programming very early computers (50s) for Operations Research. They both worked at RAND corporation and involved in early "AI" research projects for military. Those AI projects of course were logic based problem solvers of 50s and 60s. But Hubert sees a problem with this work. What they do is incompatible with the philosophical tradition he belongs: "I began reading NSS's landmark papers with a mixture of excitement and fear. Perhaps Hobbes, Kant, and Husserl were right after all, and the human mind was an analytic engine. But then what about the seemingly plausible arguments of Merleau-Ponty, Heidegger, and Wittgenstein, which I had come to accept? As I read the RAND papers my excitement and fear turned to disappointment and relief." I think this was more like a cognitive dissonance than an actual contradiction. It was about choosing sides between Heidegger and maybe Descartes for him. That's why his objection sounds personal and dogmatic. So what was the big idea of Heidegger? Elephant in the room is the concept of "Dasein" (being-there), which Dreyfus think Heidegger is a genius for being the first philosopher noticing that. Dasein is an entity, with special mode of being. Only *human-beings* can be Dasein. So it's not an "object" like a table, that has properties (like in object oriented programming). It's not an equipment like a hammer (objects having methods). Human-beings, and only human-beings (definitely not LLMs, not dogs) can have this special way of being, or Existence. This idea of course has some roots in Christian Theology, as Heidegger himself. I think this is the reason of the strong opinions. A bit dogmatic. A belief that humans are fundamentally different beings. So an Equipment trying to mimic "Dasein" is categorically wrong (even impossible) in this belief system. The problem doesn't have to be Dasein itself, but you get the idea. It's either-or. If modern AI research is not flawed, their philosophy must be flawed. Since it can't be flawed, AI research must be flawed. Since research is trial and error, failures are the part of the process. But for them, each failure is an "increasing evidence".