this post was submitted on 03 May 2024
142 points (88.2% liked)
Technology
59197 readers
3261 users here now
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related content.
- Be excellent to each another!
- Mod approved content bots can post up to 10 articles per day.
- Threads asking for personal tech support may be deleted.
- Politics threads may be removed.
- No memes allowed as posts, OK to post as comments.
- Only approved bots from the list below, to ask if your bot can be added please contact us.
- Check for duplicates before posting, duplicates may be removed
Approved Bots
founded 1 year ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
I just used ChatGPT to write a 500-line Python application that syncs IP addresses from asset management tools to our vulnerability management stack. This took about 4 hours using AutoGen Studio. The code just passed QA and is moving into production next week.
https://github.com/blainemartin/R7_Shodan_Cloudflare_IP_Sync_Tool
Tell me again how LLMs are useless?
To be honest… that doesn’t sound like a heavy lift at all.
Dream of tech bosses everywhere. Pay an intermediate dev for average level senior output.
Intermediate? Nah, junior. They're cheaper after all.
But senior devs do a lot more than output code. Sometimes - like Bill Atkinson's famous -2000 line change to Quickdraw - their jobs involve a lot of complex logic and very little actual code output.
It's a shortcut for experience, but you lose a lot of the tools you get with experience. If I were early in my career I'd be very hesitant relying on it as its a fragile ecosystem right now that might disappear, in the same way that you want to avoid tying your skills to a single companies product. In my workflow it slows me down because the answers I get are often average or wrong, it's never "I'd never thought of doing it that way!" levels of amazing.
You used the right tool for the job, saved you from hours of work. General AI is still a very long ways off and people expecting the current models to behave like one are foolish.
Are they useless? For writing code, no. Most other tasks yes, or worse as they will be confiently wrong about what you ask them.
I think the reason they're useful for writing code is that there's a third party - the parser or compiler - that checks their work. I've used LLMs to write code as well, and it didn't always get me something that worked but I was easily able to catch the error.
Only if you believe most Lemmy commenters. They are convinced you can only use them to write highly shitty and broken code and nothing else.
This is my expirence with LLMs, I have gotten it to write me code that can at best be used as a scaffold. I personally do not find much use for them as you functionally have to proofread everything they do. All it does change the work load from a creative process to a review process.
I don't agree. Just a couple of days ago I went to write a function to do something sort of confusing to think about. By the name of the function, copilot suggested the entire contents of the function and it worked fine. I consider this removing a bit of drudgery from my day, as this function was a small part of the problem I needed to solve. It actually allowed me to stay more focused on the bigger picture, which I consider the creative part. If I were a painter and my brush suddenly did certain techniques better, I'd feel more able to be creative, not less.
I would argue that there just isn't much gain in terms of speed of delivery, because you have to proofread the output - not doing it is irresponsible and unprofessional.
I don't tend to spend much time on a single function, but I can remember a time recently where I spent two hours writing a single function. I had to mentally run all cases to check that it worked, but I would have had to do it with LLM output anyway. And I feel like reviewing code is just much harder to do right than to write it right.
In my case, LLMs might have saved some time, but training the complexity muscle has value in itself. It's pretty formative and there are certain things I would do differently now after going through this. Most notably, in that case: fix my data format upfront to avoid edge cases altogether and save myself some hard thinking.
I do see the value proposition of IDEs generating things like constructors, and sometimes use such features, but reviewing the output is mentally exhausting, and it's necessary because even non-LLM sometimes comes out as broken. Assuming that it worked 100% of the time: still not convinced it amounts to much time saved at the end of day.
But we never have proofs that it gives good code, that's convenient...
So you want me to go into one of my codebases, remember what came from copilot and then paste it here? Lol no
Of course you can't.
You already forgot, that's convenient, again.
Yeah you post your employer first, dumbass
All you want is something to belittle
You say it's magical but never post proof. That's all I need to think it's shit. No need to debate about it for hours. Come back when you entice us with something instead of the billion REST APIs that are useless but seem to give a hard on to all the AI bros out there.
No one cares that you're mad about lLmS
Are you 12? What kind of answer is this?
Asking for proof for a claim is a very sane thing to do.
🤡
This is not really a slam dunk argument.
First off, this is not the kind of code I write on my end, and I don't think I'm the only one not writing scripts all day. There's a need for scripts at times in my line of work but I spend more of my time thinking about data structures, domain modelling and code architecture, and I have to think about performance as well. Might explain my bad experience with LLMs in the past.
I have actually written similar scripts in comparable amounts of times (a day for a working proof of concept that could have gone to production as-is) without LLMs. My use case was to parse JSON crash reports from a provider (undisclosable due to NDAs) to serialize it to our my company's binary format. A significant portion of that time was spent on deciding what I cared about and what JSON fields I should ignore. I could have used ChatGPT to find the command line flags for my Docker container but it didn't exist back then, and Google helped me just fine.
Assuming you had to guide the LLM throughout the process, this is not something that sounds very appealing to me. I'd rather spend time improving on my programming skills than waste that time teaching the machine stuff, even for marginal improvements in terms of speed of delivery (assuming there would be some, which I just am not convinced is the case).
On another note...
There's no need for snark, just detailing your experience with the tool serves your point better than antagonizing your audience. Your post is not enough to convince me this is useful (because the answers I've gotten from ChatGPT have been unhelpful 80% of the time), but it was enough to get me to look into AutoGen Studio which I didn't know about!
I don't think LLMs are useless, but I do think little SoC boxes running a single application that will vaguely improve your life with loosely defined AI features are useless.
Who's going to tell them that "QA" just ran the code through the same AI model and it came back "Looks Good".
:-)
The code is bad and I would not approve this. I don’t know how you think it’s a good example for LLMs.
The code looks like any other Python code out there.
We're doomed then because I would reject that in a MR for being unprofessional and full of bugs.
What bug have you spotted?
In one of those weird return None combination. Also I don’t get why it insists on using try catch all the time. Last but not least, it should have been one script only with sub commands using argparse, that way you could refactor most of the code.
Also weird license, overly complicated code, not handling HTTPS properly, passwords in ENV variables, not handling errors, a strange retry mechanism (copy pasted I guess).
It’s like a bad hack written in a hurry, or something a junior would write. Something that should never be used in production. My other gripe is that OP didn’t learn anything and wasted his time. Next time he’ll do that again and won’t improve. It’s good if he’s doing that alone, but in a company I would have to fix all this and it’s really annoying.
It's no sense trying to explain to people like this. Their eyes glaze over when they hear Autogen, agents, Crew ai, RAG, Opus... To them, generative AI is nothing more than the free version of chatgpt from a year ago, they've not kept up with the advancements, so they argue from a point in the distant past. The future will be hitting them upside the head soon enough and they will be the ones complaining that nobody told them what was comming.
Thing is, if you want to sell the tech, it has to work, and what most people have seen by now is not really convincing (hence the copious amount of downvotes you've received).
You guys sound like fucking cryptobros, which will totally replace fiat currency next year. Trust me bro.
Downvotes by a few uneducated people mean nothing. The tools are already there. You are free to use them and think about this for yourself. I'm not even talking about what will be here in the future. There is some really great stuff right now. Even if doing some very simple setup is too daunting for you, you can just watch people on youtube doing it to see what is available. People in this thread have literally already told you what to type into your search box.
In the early 90s, people exactly like you would go on and on about how stupid the computerbros were for thinking anyone would ever use this new stupid "intertnet" thing. You do you, it is totally fine if you think because a handful of uneducated, vocal people on the internet agree with you that technology has mysteriously frozen for the first time in history, then you must all be right.
If everybody in society "votes" that kind of stuff "down", the hype will eventually die down and, once the dust has settled, we'll see what this is really useful for. Right now, it can't even do fucking chatbots right (see the Air Canada debacle with their AI chatbot).
Not every invention is as significant as the Internet. There's thing like crypto which are the butt of every joke in the tech community, and people peddling that shit are mocked by everyone.
I honestly don't buy that we're on the edge of a new revolution, or that LLMs are close to true AGI. Techbros have been pushing a lot of shit that is not in alignment with regular folks' needs for the past 10 years, and have maintained tech alive artificially without interest from the general population because of venture capital.
However, in the case of LLMs, the tech is interesting and is already delivering modest value. I'll keep an eye on it because I see a modest future for it, but it just might not be as culturally significant as you think it may be.
With all that said, one thing I will definitely not do is spend any time setting up things locally, or running a LLM on my machine or pay any money. I don't think this gives a competitive edge to any software engineer yet, and I'm not interested in becoming an early adopter of the tech given the mediocre results I've seen so far.
They aren't trying to have a conversation, they're trying to convince themselves that the things they don't understand are bad and they're making the right choice by not using it. They'll be the boomers that needed millennials to send emails for them. Been through that so I just pretend I don't understand AI. I feel bad for the zoomers and genas that will be running AI and futilely trying to explain how easy it is. Its been a solid 150 years of extremely rapid invention and innovation of disruptive technology. But THIS is the one that actually won't be disruptive.
I'm not trying to convince myself of anything. I was very happy to try LLM tools for myself. They just proved to be completely useless. And there's a limit to what I'm going to do to try out things that just don't seem to work at all. Paying a ton of money to a company to use disproportionate amounts of energy for uncertain results is not one of them.
Some people have misplaced confidence with generated code because it gets them places they wouldn't be able to reach without the crutches. But if you do things right and review the output of those tools (assuming it worked more often), then the value proposition is much less appealing... Reviewing code is very hard and mentally exhausting.
And look, we don't all do CRUD apps or scripts all day.
Tell me about how when you used Llama 3 with Autogen locally, and how in the world you managed to pay a large company to use disproportionate amounts of energy for it. You clearly have no idea what is going on on the edge of this tech. You think that because you made an openai account that now you know everything that's going on. You sound like an AOL user in the 90 that thinks the internet has no real use.
I don't care about the edge of that tech. I'm not interested in investing any time making it work. This is your problem. I need a product I can use as a consumer. Which doesn't exist, and may never exist because the core of the tech alone is unsound.
You guys make grandiloquent claims that this will automate software engineering and be everywhere more generally. Show us proof. What we've seen so far is ChatGPT (lol), Air Canada's failures to create working AI chatbots (lol), a creepy plushie and now this shitty device. Skepticism is rationalism in this case.
Maybe this will change one day? IDK. All I've been saying is that it's not ready yet from what I've seen (prove me wrong with concrete examples in the software engineering domain) and given that it tends to invent stuff that just doesn't exist, it's unreliable. If it succeeds, LLMs will be part of a whole delivering value.
You guys sound like Jehovah's witnesses. get a hold of yourselves if you want to be taken seriously. All I see here is hyperbole from tech bros without any proof.
You're just saying that you will only taste free garbage wine, and nobody can convince you that expensive wine could ever taste good. That's fine, you'll just be surprised when the good wine gets cheap enough for you to afford or free. Your unwillingness to taste it has nothing to do with what already exists. In this case, it's especially naive since you could just go watch videos of people using actually good wine.
Show me proof or shut up. It's that simple. This is not a subjective matter like wine tasting. There needs to be objective and tangible proof it works.
Hyperbole again.
There are endless examples if you just search the things we've been mentioning. Here is a video that just came out today about a new project for making front ends called OpenUI.
I'll check OpenUI, thanks for the suggestion
No problem. Here is a fairly short video showing the Praison system of agents and tools. He shows how to do it locally as well, so there is no need to even use an openai api or any other remote model. This means it is simple to do the whole thing even with uncensored models or any other fine-tuned model for special use cases.
https://youtube.com/watch?v=JSU2Rndh06c
Here is an alternative Piped link(s):
https://piped.video/watch?v=JSU2Rndh06c
Piped is a privacy-respecting open-source alternative frontend to YouTube.
I'm open-source; check me out at GitHub.
Here is an alternative Piped link(s):
OpenUI
Piped is a privacy-respecting open-source alternative frontend to YouTube.
I'm open-source; check me out at GitHub.
Please show me good code done with AI. I'm waiting.
Wonderfully said, this is a very good point.