RU version is available. Content is displayed in original English for accuracy.
Advertisement
Advertisement
⚡ Community Insights
Discussion Sentiment
69% Positive
Analyzed from 1388 words in the discussion.
Trending Topics
#code#something#model#debug#why#where#models#freedom#never#need
Discussion (21 Comments)Read Original on HackerNews
LLMs generally only debug systems by reading the code with whatever information you give them in a prompt. The image in the article is meta-prompt - the prompt is whatever comes from the vision model the AI happens to use to 'understand' the red circle annotation. That won't work. To successfully debug what's going on it will need much better state information. Has the 'shelf' been explained to is? Is the contrast and lack of shadows in the image messing up the vision model? Why isn't the 'lid' in the image? And so on.
LLMs are clever but they're not magical. Treat them like a naive junior dev. Give them enough data about the state of something to understand it properly.
What I have found super useful actually is having models make a interactive 3d viewer in which I use move / highlight / paint (soft body painting directly onto the geometry for issues and different colors mean different failures). This gives a much better way to communicate the physical relationships and positions that are hard to get across in a labeled screenshot.
Its for sure still a lot of manual work so the "seeing-eye dog" description definitely holds. But I have found that after a couple of examples with the extra context the model gets much better at handling the problem and becomes useful.
Also.. uh.. isn't this.. good? I thought AI was to steal all our jobs.
___
Beside that, kinda weird self-description.
Isn't the computer executing your commands and you're just filling in where it cannot do that?
Being that dog implies that the computer is in the driver seat.
I mean it's supposed to be a joke I guess, but I read it as one that leaks internal metadata which seems to be incorrectly calibrated.
I call this "thanoscoding" for two reasons:
1. "Fine, I'll do it myself"
2. In the past I found I have to "snap away" the mess the LLM made in order to start afresh from a known good state (generally with git reset). But that was 1-2 generations ago when it comes to models. GPT6 Astra probably does things right the first time, 90% of the time.
Coding was never the reward. Acting as a translator for a machine is far worse than allowing the machine to solve the mundane parts and leave you with bigger building blocks to play with.
After 25 years I have to confess I hated being a software engineer. It felt like grinding in a video game.
Now I can finally create and innovate at the speed of thought, and I'm very grateful to have this technology now.
Giant corpos hold every sliver of your so called freedom, you dont innovate, you repeat what others created before you. You use a tool that shackles your thoughts and creativity in a never before seen way, what you perceive is an illusion of liberty that is no present. Without others you are nothing and they can take it away any seconds, you are an addict, not an innovator.
It is true that closed weights models are a big issue. It is also true that LLM-generated solutions usually drift towards a median.
But that is not the dead end you think it might be.
Most coding work is repetitive boilerplate, and most typing is just.. well.. typing. Miserable work I too did not really enjoy. What I did enjoy were the end results, and that was just a necessary step to get there.
Now that's less the case than it was before we had LLMs, and for that I too am glad.
___
I think the article headline might've primed you (and me, fwiw) to reading the comment you're replying to as passive. But if you just look at the words of it, that might not actually be the case.
I feel very much like a luddite..or some artisan of a bygone age. The guy above laments how he never enjoyed coding, this is an interesting sentiment, I know quite a few people who enjoy the craft itself, myself included. The ability to form words that have meaning, that create something from nothing. Sure, the words are just a tool, but its something _I_ can master, not some abstract wish machine that may change its functionality tommorow. Obviously one can argue that the computer itself is in this case the one that creates and not me, its not magic that just works with me, but the machine I got obeys me and me alone..another reason why personal computing is important.
Serious question - what do you do for fun? I find fishing with friends enjoyable, and using my hands tidying up the old place I bought. It’s not innovating software I’ll never use but I’ll never tire of writing the same old ASP that delivers my clients the results they’re after.
Yes, when I think of OpenAI and Anthrophic and Google and Meta and any AI labs and their intentions, i cry a single tear for how these great instituitions are working so hard to bring freedom for humanity.
AI owned by a few companies is more likely to put the majority right back to serfdom. You're a privileged fool to believe that freedom is the likely outcome of the current stampede.
With a username that appears to cheer for a sociopath¹, I doubt reason will convince you.
¹ https://futurism.com/artificial-intelligence/sources-sam-alt...
(Like, Meta’s intentions? lol)
Astra is already good at taking screenshots and acting on it (part of the agi claims).