badrunaway

6 ms·

1.
by badrunaway 2y ago
we are talking about lean proofs. Given a formal statement and a proof - the lean can verify whether it's correct or not. It's like generating computer programs to solve a problem - the problem lied in generating useful solutions&
2.
by badrunaway 2y ago
I think what we need to look at is that AI systems were not able to do it before and now they are able to do it. Sooner these systems with millions of dollars of compute will just scale the algorithms and beat humans at this as well - just
3.
by badrunaway 2y ago
This will in a few months change everything forever. Exponential growth incoming soon from Deepmind systems.
4.
by badrunaway 2y ago
I guess I need to tell them to not even send any personal detail / bank detail to anyone including me. I sort of felt like may be there is a better solution than just trusting them to follow this. Like we have insurance for death kind
5.
by badrunaway 2y ago
That's also a fear I have. But having grown up on internet it helps in adjusting to these changes fast. Secondly, I got a degree in CS and been a developer for a decade so yeah I know a video can be a deepfake instinctively. It's
6.
by badrunaway 2y ago
So, my reasoning is little stupid but it goes like this. I am in my 30s. Let's say if I lose all my money in a scam also - I have time to earn it back and still make it back. Whereas our elders (parents are 60+) who have savings and re
7.
by badrunaway 2y ago
Recently someone I know got a call where they mimicked their child's voice using AI and asked them to send immediate money otherwise child will be in danger. This has got me into thinking - my parents would also do the same for me&#x2F
8.
Ask HN: How to protect old parents from evolving AI based scams?
3 points by badrunaway 2y ago|12 comments
9.
Tonnet – mini experiment in colors and sound(nncoders.com)
1 points by badrunaway 2y ago|0 comments
10.
by badrunaway 2y ago
LLMs right now don't employ any logic. There can always be corners of "I don't know" or "I can't do that" - than the current system which is 100% confident in it's answer because it's not actuall
11.
by badrunaway 2y ago
Yes, that's why there was no human in the loop and I was using LLMs as a proxy to bottom up approach in step 1. But the hallucinations can creep into the knowledge graph also as mentioned by another commentator
12.
by badrunaway 2y ago
I thought addition of the "logical" constraints in the existing training loop using KGs and logical validation would help into reducing wrong semantic formation at the training loop itself. But your point is right that what if the
13.
by badrunaway 2y ago
Oh, exactly. But let me know your thoughts on this - let's say if you have a graph which represents existing sci-fi novel = rather than the current model which is just blindly generating text on statistical probabilities would it not h
14.
by badrunaway 2y ago
Haha, asking chat-gpt surely won't work. Everything can "feel" like a halting problem if you want perfect results with zero error with uncertain and ambiguous new data adding. My take - Hallucinations can never be made to pe
15.
by badrunaway 2y ago
I had KGs in mind with automated construction which can improve and converge during training phase. I was hypothesizing that if we give incentive during training phase to also construct KGs and bootstrap the initial KGs from existing LLMs -
16.
by badrunaway 2y ago
Interesting. I haven't really seen much into this space. But anything which can provably represent concepts and relationships without losing information can work. Devil might be in details; nothing is as simple as it looks on first sig
17.
by badrunaway 2y ago
Current architecture of LLMs focus mainly on the retrieval part and the weights learned are just converged to get best outcome for next token prediction. Whereas, ability to put this data into a logical system should also have been a train
18.
by badrunaway 2y ago
Got it. 'Finding Fun' is one good metric. I have actually a lot of free time and I want to build things that people love.
19.
by badrunaway 2y ago
yea - but that won't satisfy me. Upvote is sort of one metric on how much novel your work is? something like that.
20.
Ask HN: How to ideate small projects which HN community find fun and worthy?
2 points by badrunaway 2y ago|4 comments
21.
by badrunaway 2y ago
I believe language allows having more than one layer i.e. enabling complexity of representation. For example, If I want to think about the person thinking inside - these kind of recursive concepts need some sort of symbolic nesting to even
22.
by badrunaway 2y ago
Intuitively I feel efficiency is the outcome of existing world model.. approach can look like yours - I don't see why there has not been efforts on scaling monte carlo tree search for extending the existing world model via tree search.
23.
by badrunaway 2y ago
like darwin selection between the subsystem approaches? Put in a lot of different kind of LLMs and let them play the same game inside with each other.. whosoever wins the simulation is allowed to externally present the approach... something
24.
by badrunaway 2y ago
When we talk about system 2; is it possible that [generating large number of programs; evaluating them of the task; choosing top K outcomes; feeding it back to Neural net] can act as system 2 for a AGI? Isn't that how we think intellig
25.
by badrunaway 2y ago
Can someone summarise what this changes overall in the field? What I got is that there was no proper formalism to study emergence earlier and now we have a mathematical framework.
26.
by badrunaway 2y ago
Oh, that one I have added in code manually (500ms). Some random delay to make it sound pleasing. Thank you.
27.
by badrunaway 2y ago
I believe there should be a clear indicator in UI of every OS when any new program listens to your keystrokes.. it should be the norm
28.
by badrunaway 2y ago
nobody ported python to wasm yet?
29.
by badrunaway 2y ago
A cleaner API you mean which exposes what is necessary only.
30.
by badrunaway 2y ago
what can be done to stop all this? We need some sort of OS level layer to validate these things. If we put a local LLM which checks the bytecode of things which are getting installed/running for security = will that solve all this? My

More ›