sgt101

14 ms·

181.
by sgt101 1y ago
Humans are humans - to deny that we are thinking, reasoning, living beings is a strange thing to do. You can taste a beer, laugh so much it hurts, come to know how something works.
182.
by sgt101 1y ago
Well - all of them. The mechanism by which they work prohibits reasoning. This is easy to see if you look at a transformer architecture and think through what each step is doing. The amazing thing is that they produce coherent speech, but t
183.
by sgt101 1y ago
Thank goodness we have version control systems then.
184.
by sgt101 1y ago
LLMs are deterministic. If you run an LLM with optimization turned on on a NVIDIA GPU then you can get non-deterministic results. But, this is a choice.
185.
by sgt101 1y ago
Of course the fundamental failure case is solving novel problems. If it's in a book AI can do it. Asking if it's in the book, maybe not. Not in the book, definitely not.
186.
Things that AI cannot do(mcsweeneys.net)
39 points by sgt101 1y ago|9 comments
187.
by sgt101 1y ago
I totally agree if the chatbots are actually reliable and controllable. LLMs have real problems with both of those things, they are unreliable in the sense that they produce text with factual inaccuracies and they are uncontrollable in that
188.
by sgt101 1y ago
You are a sensible person and took professional advice. Other people are not sensible - look at all the junkies shooting who knows what into who knows where. A system that can advise cooks to put glue on pizza is probably not the best mecha
189.
by sgt101 1y ago
depends what you mean by "machine".
190.
by sgt101 1y ago
There are lots and lots of things that we can't personally observe about the universe. For example, it's quite possible that everyone in New York is holding their breath at the moment. I can't prove that either way, or determ
191.
by sgt101 1y ago
>If that is true, you have a proof that the Church-Turing thesis is false. We're talking the physical version right? I don't have any counter examples that I can describe, but I could hold that that's because human languag
192.
by sgt101 1y ago
Compute functions != Intelligence though. For example learning from experience (which LLMs cannot do because they cannot experience anything and they cannot learn) is clearly an attribute of an intelligent machine. LLMs can tell you about t
193.
by sgt101 1y ago
Ok - where do AIs put the information that they "seek" from the internet?
194.
by sgt101 1y ago
This is gold dust. Thank you for taking the time to share your knowledge.
195.
by sgt101 1y ago
Great reference - thanks.
196.
by sgt101 1y ago
ELI5: how (specifically) do GPU and TPU optimisations effect determinism in LLMs? Or is this all a myth?
197.
by sgt101 1y ago
>He said that China and the world don't see the UK as a big player anymore. Which is true. So what? You can see people however you want, but you still have to live with them in your world and they have a vote on what happens next. T
198.
by sgt101 1y ago
Well, a Marxist Leninist analyses the global scene and concludes that communism is inevitable. We've been here before.
199.
by sgt101 1y ago
Has anyone been caught by this? As in sued or prevented from working?
200.
by sgt101 1y ago
Also, books, books are really good for finding knowledge ! Seriously LLM's as a cultural technology cast them as a super interactive indexing system. I find that's a useful lens to use to understand this kind of study.
201.
by sgt101 1y ago
Is there any plausible connection between these statistics and the ending of atmospheric bomb testing? Also the various catastrophic releases of plutonium into the atmosphere from satellite reentry? Could the gradual reduction be attributed
202.
by sgt101 1y ago
>This is a deliberate choice on the part of the model makers, because a fixed checkpoint is useful for a product. They could just keep the training mechanism going, but that's like writing code without version control. Training more
203.
by sgt101 1y ago
>The diversity is itself indicative, though, that intelligence isn't bound to the particularities of the human nervous system. Across different animal species, nervous systems show a radical diversity. Different architectures; diffe
204.
by sgt101 1y ago
>which is in many ways similar to our subconscious thinking this is just made up. - we don't have any useful insight on human subconscious thinking. - we don't have any useful insight on the structures that support human subco
205.
AI Lied to Sky News Journalist (UK)(youtube.com)
1 points by sgt101 1y ago|1 comments
206.
by sgt101 1y ago
A very useful demonstration of how non technical people interact with AI and how that interaction can collapse catastrophically.
207.
by sgt101 1y ago
Super interesting. I will have to read it - I will be looking to figure out if the tasks that they are working on significant/realistic? And are the improvements that they are finding robust?
208.
by sgt101 1y ago
I was trying to pick n-shot examples from a data set. The idea was that given 1000s of examples for a prompt finding a combination of n that was optimal could be advantageous, but for n's that are large then bruteforcing the combincati
209.
by sgt101 1y ago
My brother, you haven't drunk the right coolaid. It's not impressive because text is a shit interface for interacting with a machine. GUI's are much better and we established this in like 1982. Also what they are doing in tha
210.
by sgt101 1y ago
I spent a lot of time last summer trying to get prompts to optimise using various techniques and I found that the search space was just too big to make real progress. Sure - I found a few little improvements in various iterations, but actua

More ›