General statistics
List of Youtube channels
Youtube commenter search
Distinguished comments
About
clray123
Lex Fridman
comments
Comments by "clray123" (@clray123) on "Lex Fridman" channel.
Previous
1
Next
...
All
Let me put it that way, Lex is not the brightest bulb out there, in fact rather far from it. He's just using this annoying trick of making himself sound profound by talking slowly, stressing how insightful and "empathetic" he is (an arrogant, preachy sort of fake humility), and surrounding himself by smart people. But if you asked him to accurately reproduce the trains of thought of these other people he interviews, I believe he would fail most of the time.
4
@David Eire Well, by that same logic atoms are alive because they tend to move around as well, and any chemical process involves changes in structure of matter, so it's alive as well. This is just nonsense, but the man is too famous for somebody to point it out.
2
This guy sounds like a puppet. Who is really running the OpenAI show?
2
The whole "existential" talk reeks of philosophical gobbledygook. As Lex rightly asked near the beginning, who is there to say where the drop of oil begins and where it ends. Who defines a "drop of oil"? Well, it's the modeller, and his is just one of many (perhaps infinitely many) possible models filtered through the perception of what was already in his head. Look up the term "mind projection fallacy" (coined by the physicist Jaynes, who was fond of slapping on the hands of colleagues who committed it) - as it very much seems to be applicable here... The man also mentions "probabilities", which is also a hugely problematic term upon which to build the theory of everything because of sheer arbitrariness of priors (as any Bayesian worth their salt would agree).
2
Means that his PR campaign is working.
1
SoupSandwichh He should keep his hobbies private like any normal human would. I remember he did the aqua-whatever on the lake earlier for the show, the guy's just an indolent dickhead.
1
Parts of this discussion concerning GPT-4 and AGI are just silly. 1. Clearly limited context window which lets it forget about the start of conversation in the midst of it 2. Inability to continuously learn using new information provided to it (learn as in, incorporate the information into the permanent memory and utilize in further reasoning) Does THAT sound to you like a conscious AGI, Mr. Fridman, or are you just playing stupid? (Actually, I suspect it's the latter, but also an annoying aspect of these podcasts.)
1
32:45 There is no such thing as "harmful output" for a language model, just like there is no such thing as "harmful output" from a human writer. Resorting to such language is already a big mistake toward totalitarian thinking. The proper way of phrasing it is of course that the output (just like writer's prose) may lead to harmful actions of some readers. That's the honest, civilized way to talk about these topics. Talking about "harmful output" is essentially the same as talking about "thought crime". And Mr. Lex Fridman falls for it.
1
You can see that the interviewee is embarrassingly smarter than the interviewer.
1
I'm pretty sure that in 50 years we'll have folded and unfolded all the proteins in the universe three times over while still haven't found a cure for a runny nose.
1
Whaaat? They define life by autonomous movement? What sort of bollocks is that? How about "ability to reproduce", elite thinkers?
1
@David Eire No, it's not. Viruses, for example, can reproduce without any means for autonomous movement.
1
@Dan Ellwein Mules very much try to reproduce, just fail at it.
1
The gayness has reached unprecedented levels.
1
Previous
1
Next
...
All