This article is overly dismissive, to say the least. To say the most, it's overconfident, riddled with unsupported logical premises which are never backed by an

This article is overly dismissive, to say the least. To say the most, it's overconfident, riddled with unsupported logical premises which are never backed by any concrete evidence, and a word salad. However, in the interest of not being just another jerk on the internet, I'll try to summarize my thoughts on the author's arguments in a (hopefully) somewhat constructive, concise way. I read the entire article. There's a lot of ten-dollar words, a bunch of hyphenated proxy terms, and an obvious passion behind it. However, the author's entire argument appears to boil down to "a superhuman AI can never exist because it would need to have a conscious experience above that of humans, which it can't ever do because several famous philosophers have said so." I don't want to mock, but no logical argument is presented, only the appearance of one. Statements like "in order for thing ABC to happen, XYZ must be true" are everywhere in the text, and are never proven. The author's entire case is predicated on baseless assumptions. The author says that it is impossible to replicate the "worldlike experience" that a human child goes through using things like internet datasets or physical robots with digital sensors. I actually agree with this. What I do not agree with is the logical leap from this point to "thus, a human or superhuman level AI cannot exist". No justification for this leap is given. If the robot passes the Turing test, that's all that matters. An adversarial, multi-day, modern version of a Turing test could include things like texting, calling, live video chats, etc -- basically the exact same level of communication I have with my closest friends who all live 6 hours away from me in another city. When I interact with those friends, I am convinced of their sentience. Therefore, if an AI could replicate those means of communication, then I will be convinced of its sentience. Enough of this "Something it is like to be"-ness waffling or overcomplicated thought experiments. We already have a very easy, very strictly defined concept of sentience in the form of the Turing test, and it is becoming increasingly obvious that SOTA models are trending towards beating it. Once they beat it, they are for all intents and purposes alive, thinking, sentient, whatever you want to call it. They are indistinguishable from us. From that point it's really not hard to imagine superhuman models. Just imagine what you would do if you had the ability to instantly scan through gigabytes of text in a database or predict the behavior of every other person around you eerily well or perform arithmetic at clearly superhuman speeds. It's really not that much of a stretch to see superhuman AIs as just digital humans with fantastic brain-computer-interfaces. Anyways, those are my (admittedly long-winded, less-than-constructive) two cents.