johnthewise

8 ms·

61.
by johnthewise 2y ago
>There really is something to sentience that we all know that cannot be reduced to language or machinery. >We are complete beings - my intelligence is part of a system, and it makes no sense to abstract it out of that system in order
62.
by johnthewise 2y ago
sorry, my bad. I assumed GPT-4 compute to be around 100M-500M, so 80X compute of gpt4 would be around 10B-40B.
63.
by johnthewise 2y ago
It's not just him though. His views are very common in AI labs, who actually build these systems and have better visibility than the rest of us on what's going on.
64.
by johnthewise 2y ago
>Talk amongst the memesters. This is a meme. Those memesters invented and deployed the current systems, are currently training the next, and raising funds for the next after. Some scientific breakthroughs were memes at first, but some pe
65.
by johnthewise 2y ago
>That's... probably not sustainable for four more years. If it is, it's not sustainable much past that. You're probably not going to see a ten-trillion-dollar cluster; you're definitely not going to see a one-hundred-
66.
by johnthewise 2y ago
Some people can't do it whether they had the time to study, which the point of the test.
67.
by johnthewise 2y ago
Supposedly that was Q* all about. Search recursively, backtrack if dead end. who knows really, but the technology is still very new, I personally don't see why a sufficiently good world model can't be used in this manner.
68.
by johnthewise 2y ago
This type of argument can be used against pretty much any group that has ever existed, then either it's too broad and not a meaningful critique of the group itself or a critique of humans getting together in general.
69.
by johnthewise 3y ago
How could there be no innovation across thousands of years? It's not like we suddenly decided to innovate at BC7380, some people somewhere at all times must have kept innovating, at least to beat other tribes, farm/hunt better.
70.
by johnthewise 3y ago
Claude opus is better in my experience
71.
by johnthewise 3y ago
There is no single definition of AGI. Performing most intellectual tasks human perform today is both general and a form of intelligence, so I too agree with it.
72.
by johnthewise 3y ago
If you hate your job. If you love it, it's play time.
73.
by johnthewise 3y ago
I like the example of dotcom bubble enabling fiber investments. Hyping up internet applications enabled setting up infrastructure, which in turn made the hype real, albeit years later. I think in this wave, GPUs will be largely outdated in
74.
by johnthewise 3y ago
Your curiosity is a bit of fresh air after months of seeing people arguing over pointless semantics. So I'm going to attempt to explain my mental model of how this works. 1- This is correct but not really useful view imo. Saying it can
75.
by johnthewise 3y ago
Tesla is making the world a better place by pushing EV adoption around the world. Have you done anything better for the world today, yet alone for its future? Starlink is pretty neat as well, he could fail going to Mars and all that and he
76.
by johnthewise 3y ago
there doesn't have to be moat to provide some value to some people. These are mostly indie hackers trying to build a nice UI/UX on ai workflows to actually make them usable by general population. Even with retrieval integration of
77.
by johnthewise 3y ago
It doesn't even need to hack or do anything illegal. It can make money trading stocks, predicting or modelling far better than market participants, use money to influence the real world, bribe politicians etc. Like with few billions, i
78.
by johnthewise 3y ago
yeah, he proposed bombing rogue data centers by enforcing countries to be a viable option. he proposes this so we have a chance of solving alignment first, if that's even possible, not eliminating this risk once and for all. In the lim
79.
by johnthewise 3y ago
He is not really convinced you can align an AI that is intelligent past some threshold with aims of a single entity. His risk calculations derive primarily from the AI itself, not weaponization of AI by others. Rich Sutton seems to agree wi
80.
by johnthewise 3y ago
Which foundational model did you finetune with few images? just curious. I personally believe language interface or language conditioning is not very relevant or even harmful for many downstream CV applications. In your case, you don't
81.
by johnthewise 3y ago
GPT4 being a mixture of experts is irrelevant imo like we don't care about how many layers there are in a network and how wide those layers are or which type of activation functions are actually used etc. all that matters are we can ru
82.
by johnthewise 3y ago
>AI can't act on its own in the physical world. If it can earn money or is given money, it can certainly get people to do things in the physical world.
83.
by johnthewise 3y ago
You are right, but evidence of LLMs working well with language is not a sign that they might be memorizing Chess. But rather opposite, infeasability of memorization of chess and yet LLMs still doing well on them might be a sign that LLMs c
84.
by johnthewise 3y ago
It's not a trick. Every information can be represented by sequence of bits or strings. It's not the surprising part. It's just that its a telling sign that through compression understanding of the game emerges because: 1- unl
85.
by johnthewise 3y ago
Sure, that wouldn't be impressive if you wrote tree search for chess. However, if you wrote a program that doesn't do any search and still plays well, that'd be impressive. I don't understand your claim that LLMs playing
86.
by johnthewise 3y ago
Agreed. I find shallow skepticism to be very sad way of thinking. If the stakes are not high, i.e. you are not putting your life savings into it, why don't you allow your imagination to be entertained for a brief of time? If it turns o
87.
by johnthewise 3y ago
I like this quote from Napoleon on taking risks: “If the art of war were nothing but the art of avoiding risks, glory would become the prey of mediocre minds.... I have made all the calculations; fate will do the the rest."
88.
by johnthewise 3y ago
they work with compressed representations, you take an arbitrary information with varying entropy into a fized size vector representation, that's a compression.
89.
by johnthewise 3y ago
Constitutional part refers to set of ethical rules the agent must reflect on to iterate his response. i.e. it's not just "make it more ethical" but "basen on theseruels(constitution), can you improve your response?"
90.
How Does GPT Obtain Its Ability? Tracing Emergent Abilities of Language Models(yaofu.notion.site)
4 points by johnthewise 3y ago|0 comments

More ›