There’s a good point here that like about 80% of what we’re calling AI right now… isn’t even AI or even LLM. It’s just… algorithm, code, plain old math. I’m pretty sure someone is going to refer to a calculator as AI soon. “Wow, it knows math! Just like a person! Amazing technology!”
(That’s putting aside the very question of whether LLMs should even qualify as AIs at all.)
In my professional experience, AI seems to be just a faster way to generate an algorithm that is really hard to debug. Though I am dev-ops/sre so I am not as deep in it as the devs.
I remined of the time researchers used an evolutionary algorithm to devise a circuit that would emit a tone on certain audio inputs and not on others. They examined the resulting circuit and found an extra vestigial bit, but when they cut it off, the chip stopped working. So they re-enabled it. Then they wanted to show off their research at a panel, and at the panel it completely failed. Dismayed they brought it back to their lab to figure out why it stopped working, and it suddenly started working fine.
After a LOT of troubleshooting they eventually discovered that the circuit was generating the tone by using the extra vestigial bit as an antenna that picked up emissions from a CRT in the lab and downconverted it to the desired tone frequency. Turn of the antenna, no signal. Take the chip away from that CRT, no signal.
That’s what I expect LLMs will make. Complex, arcane spaghetti stuff that works but if you look at it funny it won’t work anymore, and nobody knows how it works at all.
As a devops person, I’m constantly jumping back and forth to whatever programming language and tools each team uses. Sometimes it takes a bit to find the context, and I’m hoping ai can help. Unfortunately, allowing the ai to see code is currently off limits by corporate policy, so it only helps in those situations where I need to generate boilerplate
In my jobs there have slways been certain stule requirements to the code. AI doesn’t take those into account. So I would have to rework the code anyway. And of course there are the local libraries it know nothing about.
Fight technology with technology. I’m sure you can specify a style for it to generate, but we already run everything through a prettifier configured for what we look for …. Unless you mean a higher order like naming or architecture
Lol, the lead can’t spec the style, he just reviews the code and asks for changes. Sometimes it’s just that we already have a method that does a similar thing, so we should use it. Of course an AI wouldn’t know about that unless you gave it access to your code. And given how speed first AI companies are, I would never trust that data with them. But other times it’s just the leads personal preference.
I just had to transfer one of my guys out after frequent arguments to do that. I don’t understand - I point out a function that does exactly why he wants, yet he still wants to reinvent it.
I’m dreading when I come back after break. I got 50% a new junior guy who keeps saying he’s a great programmer. No sign of it so far but my management insists I take him on. All he needs to do is expose a new endpoint, wire up functionality that’s already there, and I walked him through it. Should be easy, right? No reinventing the wheel, right?
I do know the feel. I was young once (seems a long long time ago). Reinventing the wheel was fun and challenging. I loved creating libraries. I didn’t actually know at the time that there were already existing libraries I could just use, this was the 90s. But it is hard to keep track of what library code exists in a project if you didn’t write it. Maybe assign them to review the library code and document it or something. Then they will know what it is at least.
There’s a good point here that like about 80% of what we’re calling AI right now… isn’t even AI or even LLM. It’s just… algorithm, code, plain old math. I’m pretty sure someone is going to refer to a calculator as AI soon. “Wow, it knows math! Just like a person! Amazing technology!”
(That’s putting aside the very question of whether LLMs should even qualify as AIs at all.)
In my professional experience, AI seems to be just a faster way to generate an algorithm that is really hard to debug. Though I am dev-ops/sre so I am not as deep in it as the devs.
I remined of the time researchers used an evolutionary algorithm to devise a circuit that would emit a tone on certain audio inputs and not on others. They examined the resulting circuit and found an extra vestigial bit, but when they cut it off, the chip stopped working. So they re-enabled it. Then they wanted to show off their research at a panel, and at the panel it completely failed. Dismayed they brought it back to their lab to figure out why it stopped working, and it suddenly started working fine.
After a LOT of troubleshooting they eventually discovered that the circuit was generating the tone by using the extra vestigial bit as an antenna that picked up emissions from a CRT in the lab and downconverted it to the desired tone frequency. Turn of the antenna, no signal. Take the chip away from that CRT, no signal.
That’s what I expect LLMs will make. Complex, arcane spaghetti stuff that works but if you look at it funny it won’t work anymore, and nobody knows how it works at all.
As a devops person, I’m constantly jumping back and forth to whatever programming language and tools each team uses. Sometimes it takes a bit to find the context, and I’m hoping ai can help. Unfortunately, allowing the ai to see code is currently off limits by corporate policy, so it only helps in those situations where I need to generate boilerplate
In my jobs there have slways been certain stule requirements to the code. AI doesn’t take those into account. So I would have to rework the code anyway. And of course there are the local libraries it know nothing about.
Fight technology with technology. I’m sure you can specify a style for it to generate, but we already run everything through a prettifier configured for what we look for …. Unless you mean a higher order like naming or architecture
Lol, the lead can’t spec the style, he just reviews the code and asks for changes. Sometimes it’s just that we already have a method that does a similar thing, so we should use it. Of course an AI wouldn’t know about that unless you gave it access to your code. And given how speed first AI companies are, I would never trust that data with them. But other times it’s just the leads personal preference.
I just had to transfer one of my guys out after frequent arguments to do that. I don’t understand - I point out a function that does exactly why he wants, yet he still wants to reinvent it.
I’m dreading when I come back after break. I got 50% a new junior guy who keeps saying he’s a great programmer. No sign of it so far but my management insists I take him on. All he needs to do is expose a new endpoint, wire up functionality that’s already there, and I walked him through it. Should be easy, right? No reinventing the wheel, right?
I do know the feel. I was young once (seems a long long time ago). Reinventing the wheel was fun and challenging. I loved creating libraries. I didn’t actually know at the time that there were already existing libraries I could just use, this was the 90s. But it is hard to keep track of what library code exists in a project if you didn’t write it. Maybe assign them to review the library code and document it or something. Then they will know what it is at least.