I work as a programmer and I didn't realize how many people have already adopted AI into their workflow. About half of my coworkers (most younger people) ask chatGPT to write code for whatever they need to program before starting. Even after corporate emails about not sharing IP and trade secrets with AI people still do it. AI is a powerful tool and it cannot be un-invented. People will use it as long as it continues to make their lives easier.
Please tell me you don't work with medical or aerospace/military software.
If you think those industries work on the competency of their programmers instead of the competency of their systems you've got a big surprise coming for you
I don't think my experience is standalone in this. People will use tools to help them however they can. What sector a programmer in will likely have no affect on AI use. It may be harder to use on a secure network but people can still do it on their phone or personal computer. And medical and aerospace is heavily regulated and tested so the likelihood of bad AI code getting through is very low. (not to say coding mistakes don't happen, just look at the boeing 737 max crashes) And militaries also test their equipment before/after purchase. While they are often held to different standards than commercial equipment a military usually has people who are competent reviewing code and equipment.
People will use it as long as it continues to make their lives easier.
Being fired and sued for divulging trade secrets doesn't sound like an easy life.
Using ChatGPT for coding is like copy pasting from stack overflow. useful for someone who's bad at his jobs, doesn't change shit for a good developer
It depends how you use it, I think, like any tool. I might ask ChatGPT to help me write an algorithm to do a certain thing in pseudocode so I can understand it. Ask it a few questions about optimization just so I have a sense for how it works then I implement it myself.
It can also help you think about ideas. I will copy paste a function or file and then ask questions like "what are some considerations do you think I should have?" "is there anything I could be missing?" "what could make this code better?" "how would you optimize this?" "how would you make it simpler?"
let me find some simple example
function findAbsoluteMax(arr) {
return arr.reduce((val, next) => {
if (Math.abs(next) > Math.abs(val)) {
return next;
} else {
return val;
}
});
}
Let's ask GPT4 "how could we make this code better?"
It offers two suggestions
there should be some simple error handling. for example if the arr is length 0 then it should throw an error or return a null. this makes sense and is a good thing to add - perhaps this would have saved me a lot of headache in some scenario where I'm getting a weird bug
add a ternary operator to make the arr.reduce call shorter
return arr.reduce((val, next) => Math.abs(next) > Math.abs(val) ? next : val );
I think this does actually make it more readable and condenses it - a pretty good thing
Now, this is a simple function but you can actually copy in a whole file and ask it to analyze things you might be missing or considerations you could make. It's like talking to the yellow duck except the yellow duck talks back
There's a lot of power in this technology and it doesn't simply revolve around copy pasting code. Perhaps my example wasn't the greatest but someone else can share how they use it
Yea I pretty much use ChatGPT as an interactive rubber duck when working. However I’ve refrained from pasting a file or a code snippet from my work to it as there’s already some IP leaks happened at Samsung and my company has shared a guideline on how to use AI tools to help with work. They know that people will keep using it regardless, so what they can do is to keep people from leaking company or client IPs by sharing a file to the AI tools.
well that sounds like…
😀 🕶️ 😎
good news
Tell me how you'd enforce it?
It's easy enough to wash a generated text and AI text catchers don't work.
Is it really just an honor system?
Always has been.
The way you enforce any rule.
So, how do you enforce any rule?
Define a rule, announce it to people, educate them if needed on the subject, monitor and reward those, who obey the rule, establish consequences for violating it, establish a reporting mechanism, evaluate and improve at any step if something is not working.
ChatGPT is unreliable, but AIs that can search the internet can be just as reliable and trustworthy as human authors. Of course, Bing Chat is not FOSS, so I don't fully support it, but it is very good at writing accurate articles.
"just as trustworthy as human authors" - Ok so you have no idea how these chatbots work do you?
You have a lot of faith in human authors.
Oh I do not, but the choice is: a human who might understand what happens vs a probabilistic model that is unable to understand ANYTHING
LLM AI bases its responses from aggregated texts written by ... human authors, just without having any sense of context or logic or understanding of the actual words being put together.
I understand they are just fancy text prediction algorithms, which is probably justa as much as you do (if you are a machine learning expert, I do apologise). Still, the good ones that get their data from the internet rarely make mistakes.
I'm not an ML expert but we've been using them for a while in neurosciences (software dev in bioinformatics). They are impressive, but have no semantics, no logics. It's just a fancy mirror. That's why, for example, world of warcraft player have been able to trick those bots into making an article about a feature that doesn't exist.
Do you really want to lose your time reading a blob of data with no coherency?
Do you really want to lose your time reading a blob of data with no coherency?
We are both on the internet, lol. And I mean it. LLMs are slightly worse than the CEO-optimized clickbaity word salad you get in most articles. Before you've found out how\where to search for direct and correct answers, it would be just the same or maybe worse. <– I found this skill a bit fascinating, that we learn to read patterns and red flags without even opening a page. I doubt it's possible to make a reliable model with that bullshit detector.
ChatGPT can also search the internet
Didn’t they turned off that feature? Or has it been turned back on now?
I don’t think it was ever turned off, it just requires the subscription to access GPT-4 and then enabling the plugins.
It was a closed beta before, but it’s been available to everybody for a while now.
There was also the version with Bing integration that they removed, which might be what you’re thinking of… but there are 100’s of other web search plugins available beyond Bing.
This is a most excellent place for technology news and articles.
