golol

35 ms·

421.
by golol 3y ago
I disagree. It is like 80% there.
422.
by golol 3y ago
In my opinion, a terrible article. If you are so much better at knowing what is profitable and what is not, why don't you invest and make a quick billion?
423.
by golol 3y ago
I didn't know that quantum decoherence tackles the measurement problem. It seems like quite an intuitive explanation and does preserve a global wavefunction. Perhaps this doesn't quite count this as "standard quantum mechanic
424.
by golol 3y ago
I don't think you can reason like this. As far as I understodod, standard quantum mechanics does not make any statement about how the measuring process and the collapse of the wavefunvtion happens. So while the waveform evolves determi
425.
by golol 3y ago
You misunderstand. He is like you. He felt anxious when he wasn't being productive so he spent his free time studying or whatever. But at some point maybe that stops working for you. Maybe you just don't have energy left and need
426.
by golol 3y ago
It is a good thing if people actually fill up the trains and get value out of them. While empty trains are comfortable, they are an inefficiency. The 9€ ticket showed that people would really like to be more mobile ´if they could afford it.
427.
by golol 3y ago
I feel like this task might be testing if LLMs can reason about ambiguity more so than if LLMs model ambiguity. To test the latter I would give it natural language tasks that are ambigious require the recognition and modelling of this ambig
428.
by golol 3y ago
I find this very confusingly written. If they just dumped a large chat log of the evaluation process it would be much clearer what the model is actually being asked to do. With such a delicate task one has to make sure that the model actual
429.
by golol 4y ago
There are more situations than you might think where the correct answer is easy to verify but hard to find. If you are building something concrete you can just check if it works. If you are trying to understand something, it can be that Cha
430.
by golol 4y ago
Hm, I wonder if the obvious faults in the picture generated by Dalle-2 were interpreted to be advanced and creative photography techniques.
431.
by golol 4y ago
> A set of additional measures concern the availability of tools to enable data subjects, including non-users, to obtain rectification of their personal data as generated incorrectly by the service, or else to have those data erased if r
432.
by golol 4y ago
Once you have an agentic AI with years of experience your claim is disproven. I consider this as conceptually possible given a 10-100x increase in scale.
433.
by golol 4y ago
> If I were to start with ChatGPT I would, necessarily, only say what has been said. Perhaps with ChatGPT you would actually go ahead and write that book, adding those crucial 10% of key ideas which are yours and letting it help you fill
434.
by golol 4y ago
What is the difference between removal and mitigation, i.e. information and intelligence here?
435.
by golol 4y ago
I try not to define what intelligence is. I just think that for any definition of intelligence, you can speed it up and you can put it in parallel (let it cooperate). And I think this is worthy of note.
436.
by golol 4y ago
I think to include this vague sense of cooperation in the concept of intelligence also arises quite naturally from the trivial form of cooperation that is parallel computing.
437.
A basic mathematical structure of intelligence(lesswrong.com)
10 points by golol 4y ago|7 comments
438.
by golol 4y ago
I have hope that AI tutors can lead to perhaps the biggest reduction in education inequality in human history, similar to the printing press.
439.
by golol 4y ago
The solution is to weigh the data by its engagement. If in a hypothetical scenario the internet is littered with 90% Ai generated trash, then human users will only be engaging with the 10% of valuable content, which may have been created by
440.
Ask HN: Is image-GPT-4 better at tikz/canvas drawings than text-only GPT-4?
1 points by golol 4y ago|1 comments
441.
by golol 4y ago
This is wrong. I asked GPT-4 right now to perform this task and it got 3/3 3-digit calculations correct, and on a 4-digit calculation was off by 10 (not 1!). And note that artihmetic is seen as a weak spot of LLMs, it is not a good exa
442.
by golol 4y ago
I used GPT-4 to write a simple chat website so I can talk to it without having to do manual API calls. It created a working version on the first try and me telling it about a few errors. In an hour I had improved it to a really polished loo
443.
by golol 4y ago
AI only has to be as reliable as a human. The task is ultimately trivial and clearly in the capability of GPT-4. It is easy to statistically verify if GPT-4 has higher or equal correctness than a human. Of course Tao did not do this here, b
444.
by golol 4y ago
It's a pretty obvious idea executed well. I definely think symbolic AI agents written in the programming language english and interpreted using LLMs is the way forward.
445.
by golol 4y ago
No. The prompt can condition the probabilities to correspond to a higher IQ output.
446.
by golol 4y ago
It is an algorithm running a computer. The software is whatever you prompt engineered. That is the algorithm running on the computer. You know, I think that some people (I see on twitter, probably not you) have a wrong intuition about art
447.
by golol 4y ago
Well all the information it learns at runtime is encoded in the context window. I don't feel like {tokens}^ctxWindow is unmeasurably complex. I think one should see a transformer as a stochastic computer operating on its memory. If you
448.
by golol 4y ago
>Then by that reasoning everything in the physical world is a Markov chain, right? Well I guess maybe it's true that you can turn any stochastic process into a Markov Chain by changing the state space somehow (for example the states
449.
by golol 4y ago
It is a Markov Chain on the state space {Tokens}^CtxWindow.
450.
by golol 4y ago
It's wrong. A decoder only transformer performs a (possibly random) operation on a state from the state space {tokens}^CtxWindow, where the distribution of the new state depends entirely on the previous state. It is a Markov Chain with

More ›