9 ms·

Also, I think we are quite a ways out from a tool being able to devise a solution to a complex high-level problem without online precedent, which is where I fin

Also, I think we are quite a ways out from a tool being able to devise a solution to a complex high-level problem without online precedent, which is where I find the most satisfaction anyway. LLMs in particular can be a very fast, surprisingly decent (but, as you mention, very fallible) replacement for Stack Overflow, and, as such, a very good complement to a programmer's skills – seems to me like a net positive at least in the near to medium term.

al_borland 3y ago
It's also where I find most of the work. There are plenty of off the shelf tools to solve all the needs of the company I work at. However, we still end up making a lot of our own stuff, because we want something that the off the shelf option doesn't do, or it can't scale to the level we need. Other times we buy two tools that can't talk to each other and need to write something to make them talk. I often hear people online say they simply copy/paste stuff together from Stack Overflow, but that has never been something I could do at my job. My concern isn't about an LLM replacing me. My concern is our CIO will think it can, firing first, and thinking later.
- cjbgkagh 3y ago
  It’s not just about if a LLM could replace you, if a LLM replaces other enough other programmers it’ll tank the market price for your skills.
  - fisherjeff 3y ago
    I think another possibility is if you have skills that an LLM can’t replicate, your value may actually increase.
    - cjbgkagh 3y ago
      Only if the other people that the LLM did replace cannot cross train into your space. Price is set at the margins. People imagine it’ll be AI taking the jobs but mostly it’ll be people competing with other people for the space that’s left after AI has taken its slice.
  - fiddlerwoaroof 3y ago
    I don’t think this will happen because we’ll just increase the complexity of the systems we imagine. I think a variant of Wirth’s law applies here: the overall difficulty of programming tasks stays constant because, when a new tool simplifies a previously hard task, we increase our ambitions.
    - cjbgkagh 3y ago
      In general people are already working at their limits, tooling can help a bit but the real limitation to handling complexity is human intelligence and that appears to be mostly innate. The people this replaces can’t exactly skill up to escape the replacement, and the AI will keep improving so the proportion being replaced will only increase. As someone near the top end of the skill level my hope is that I’ll be one of the last to go, I’ll hopefully make enough money in that time to afford a well stocked bunker.
      - fiddlerwoaroof 3y ago
        But, for example, I probably couldn’t have written a spell checker myself forty years ago. Now, something like aspell or ispell is just an of the shelf library. Similarly, I couldn’t implement Timely Stream Processing in a robust way, but flink makes it pretty easy for me to use with a minimal conceptual understanding of the moving parts. New abstractions and tools raise the floor, enabling junior and mid-level engineers to do what would have taken a much more senior engineer before they existed.
        cjbgkagh 3y ago
        I used to think that way but from my experience and observations I've found that engineers are more limited by their innate intelligence rather than their tooling. Experience counts but without sufficient intelligence some people will never figure out certain things no matter how much experience they have - I wish it wasn't so but it's the reality that I have observed. Better tooling will exacerbate the difference between smart and not so smart engineers with the smart engineers becoming more productive and the not so smart engineers will instead be replaced.
  - saurik 3y ago
    If an LLM gets good enough to come for our jobs it is likely to replace all the people who hire us, all the way up to the people who work at the VC funds that think any of our work had value in the first place (remember: the VC fund managers are yet more employees that work for capital, and are just as subject to being replaced as any low-level worker).
    - fragmede 3y ago
      that's true, but it's harder to replace someone when you have a personal connection to them. VC fund managers are more likely to be personally known to he person who signs the checks. low-level workers may never have spoken any words to them or even ever have met them.
- quantum_state 3y ago
  Then the CIO itself gets fired … after all, average per job life of a CIO is roughly 18 months
josephg 3y ago
We’ll see - but given the gap between chatgpt 3 and 4, I think AIs will be competitive with mid level programmers by the end of the decade. I’d be surprised if they aren’t. The training systems we use for LLMs are still so crude. ChatGPT has never interacted with a compiler. Imagine learning to write code by only reading (quite small!) snippets on GitHub. That’s the state llms are in now. It’s only a matter of time before someone figures out how to put a compiler in a reinforcement learning loop while training an LLM. I think the outcome of that will be something that can program orders of magnitude better. I’ll do it eventually if nobody else does it first. We also need to solve the “context” problem - but that seems tractable to me too. For all the computational resources they use to do training and inference, our LLMs are still incredibly simple. The fact they can already code so well is a very strong hint for what is to come.
- kenjackson 3y ago
  Why do you say the snippets are small? They don’t get trained on the full source files?
  - mike_hearn 3y ago
    Nope. LLMs have a limited context window partly because that's the chunk size with which they're presented with data to learn during training (and partly for computational complexity reasons). One of the reasons I'm feeling very bullish on LLMs is because if you look at the exact training process being used it's full of what feels like very obvious low hanging fruit. I suspect a part of the reason that training them is so expensive is that we do it in really dumb ways that would sound like a dystopian hell if you described it to any actual teacher. The fact that we can get such good results from such a terrible training procedure by just blasting through it with computational brute force, strongly suggests that much better results should be possible once some of that low hanging fruit starts being harvested.
- DrSiemer 3y ago
  With today's mid level programmers, yes. But by that time, many of today's mid level programmers will be able to do stuff high level programmers do today. Many people underestimate an LLM's most powerful feature when comparing it with something like Stackoverflow: the ability to ask followup questions and immediately get clarification on anything that is unclear. I wish I had had access to LLM's when I was younger. So much time wasted on repetitive, mundane in-between code...
TerrifiedMouse 3y ago
> LLMs in particular can be a very fast, surprisingly decent (but, as you mention, very fallible) replacement for Stack Overflow Nice thing about Stack Overflow is it’s self-correcting most of the time thanks to, https://xkcd.com/386/ https://xkcd.com/386/ GPT not so much.
agotterer 3y ago
Spreadsheets didn’t replace accountants, however, it made them more efficient. I don’t personally believe AI will replace software engineers anytime soon, but it’s already making us more efficient. Just as Excel experience is required to crunch numbers, I suspect AI experience will be required to write code. I use chat-gpt every day for programming and there are times where it’s spot on and more times where it’s blatantly wrong. I like to use it as a rubber duck to help me think and work through problems. But I’ve learned that whatever the output is requires as much scrutiny as a good code review. I fear there’s a lot of copy and pasting of wrong answers out there. The good news is that for now they will need real engineers to come in and clean up the mess.
- aabhay 3y ago
  Spreadsheets actually did put many accountants and “computers” (the term for people that tallied and computed numbers, ironically a fairly menial job) out of business. And it’s usually the case that disruptive technology’s benefits are not evenly distributed. In any case, the unfortunate truth is that AI as it exists today is EXPLICITLY designed to replace people. That’s a far cry from technologies such as the telephone (which by the way put thousands of Morse code telegraph operators out of business)
  - bdw5204 3y ago
    It is especially sad that VC money is currently being spent on developing AI to eliminate good jobs rather than on developing robots to eliminate bad jobs.
    - pama 3y ago
      The plan has always been to build the robots together with the better AI. Robots ended up being much harder than early technologists imagined for a myriad different reasons. It turned out that AI is easier or at least that is the hope.
      - grogenaut 3y ago
cornel_io 3y ago
Two years ago we were quite a ways out from having LLMs that could competently respond to commands without getting into garbage loops and repeating random nonsense over and over. Now nobody even talks about the Turing test anymore because it's so clearly been blown past. I wouldn't be so sure it will be very long before solving big, hard, and complex problems is within reach...
jamaicahest 3y ago
> LLMs in particular can be a very fast, surprisingly decent (but, as you mention, very fallible) replacement for Stack Overflow I think that sentence nails it. For the people who consider "searching stackoverflow and copy/pasting" as programming, LLMs will replace your job, sure. But software development is so much more, critical thinking, analysing, gathering requirements, testing ideas and figuring out which to reject, and more.

al_borland 3y ago

It's also where I find most of the work. There are plenty of off the shelf tools to solve all the needs of the company I work at. However, we still end up making a lot of our own stuff, because we want something that the off the shelf option doesn't do, or it can't scale to the level we need. Other times we buy two tools that can't talk to each other and need to write something to make them talk. I often hear people online say they simply copy/paste stuff together from Stack Overflow, but that has never been something I could do at my job. My concern isn't about an LLM replacing me. My concern is our CIO will think it can, firing first, and thinking later.

cjbgkagh 3y ago
It’s not just about if a LLM could replace you, if a LLM replaces other enough other programmers it’ll tank the market price for your skills.
- fisherjeff 3y ago
  I think another possibility is if you have skills that an LLM can’t replicate, your value may actually increase.
  - cjbgkagh 3y ago
    Only if the other people that the LLM did replace cannot cross train into your space. Price is set at the margins. People imagine it’ll be AI taking the jobs but mostly it’ll be people competing with other people for the space that’s left after AI has taken its slice.
- fiddlerwoaroof 3y ago
  I don’t think this will happen because we’ll just increase the complexity of the systems we imagine. I think a variant of Wirth’s law applies here: the overall difficulty of programming tasks stays constant because, when a new tool simplifies a previously hard task, we increase our ambitions.
  - cjbgkagh 3y ago
    In general people are already working at their limits, tooling can help a bit but the real limitation to handling complexity is human intelligence and that appears to be mostly innate. The people this replaces can’t exactly skill up to escape the replacement, and the AI will keep improving so the proportion being replaced will only increase. As someone near the top end of the skill level my hope is that I’ll be one of the last to go, I’ll hopefully make enough money in that time to afford a well stocked bunker.
    - fiddlerwoaroof 3y ago
      But, for example, I probably couldn’t have written a spell checker myself forty years ago. Now, something like aspell or ispell is just an of the shelf library. Similarly, I couldn’t implement Timely Stream Processing in a robust way, but flink makes it pretty easy for me to use with a minimal conceptual understanding of the moving parts. New abstractions and tools raise the floor, enabling junior and mid-level engineers to do what would have taken a much more senior engineer before they existed.
      - cjbgkagh 3y ago
        I used to think that way but from my experience and observations I've found that engineers are more limited by their innate intelligence rather than their tooling. Experience counts but without sufficient intelligence some people will never figure out certain things no matter how much experience they have - I wish it wasn't so but it's the reality that I have observed. Better tooling will exacerbate the difference between smart and not so smart engineers with the smart engineers becoming more productive and the not so smart engineers will instead be replaced.
- saurik 3y ago
  If an LLM gets good enough to come for our jobs it is likely to replace all the people who hire us, all the way up to the people who work at the VC funds that think any of our work had value in the first place (remember: the VC fund managers are yet more employees that work for capital, and are just as subject to being replaced as any low-level worker).
  - fragmede 3y ago
    that's true, but it's harder to replace someone when you have a personal connection to them. VC fund managers are more likely to be personally known to he person who signs the checks. low-level workers may never have spoken any words to them or even ever have met them.
quantum_state 3y ago
Then the CIO itself gets fired … after all, average per job life of a CIO is roughly 18 months

josephg 3y ago

We’ll see - but given the gap between chatgpt 3 and 4, I think AIs will be competitive with mid level programmers by the end of the decade. I’d be surprised if they aren’t. The training systems we use for LLMs are still so crude. ChatGPT has never interacted with a compiler. Imagine learning to write code by only reading (quite small!) snippets on GitHub. That’s the state llms are in now. It’s only a matter of time before someone figures out how to put a compiler in a reinforcement learning loop while training an LLM. I think the outcome of that will be something that can program orders of magnitude better. I’ll do it eventually if nobody else does it first. We also need to solve the “context” problem - but that seems tractable to me too. For all the computational resources they use to do training and inference, our LLMs are still incredibly simple. The fact they can already code so well is a very strong hint for what is to come.

kenjackson 3y ago
Why do you say the snippets are small? They don’t get trained on the full source files?
- mike_hearn 3y ago
  Nope. LLMs have a limited context window partly because that's the chunk size with which they're presented with data to learn during training (and partly for computational complexity reasons). One of the reasons I'm feeling very bullish on LLMs is because if you look at the exact training process being used it's full of what feels like very obvious low hanging fruit. I suspect a part of the reason that training them is so expensive is that we do it in really dumb ways that would sound like a dystopian hell if you described it to any actual teacher. The fact that we can get such good results from such a terrible training procedure by just blasting through it with computational brute force, strongly suggests that much better results should be possible once some of that low hanging fruit starts being harvested.
DrSiemer 3y ago
With today's mid level programmers, yes. But by that time, many of today's mid level programmers will be able to do stuff high level programmers do today. Many people underestimate an LLM's most powerful feature when comparing it with something like Stackoverflow: the ability to ask followup questions and immediately get clarification on anything that is unclear. I wish I had had access to LLM's when I was younger. So much time wasted on repetitive, mundane in-between code...

TerrifiedMouse 3y ago

> LLMs in particular can be a very fast, surprisingly decent (but, as you mention, very fallible) replacement for Stack Overflow Nice thing about Stack Overflow is it’s self-correcting most of the time thanks to, https://xkcd.com/386/ https://xkcd.com/386/ GPT not so much.

agotterer 3y ago

Spreadsheets didn’t replace accountants, however, it made them more efficient. I don’t personally believe AI will replace software engineers anytime soon, but it’s already making us more efficient. Just as Excel experience is required to crunch numbers, I suspect AI experience will be required to write code. I use chat-gpt every day for programming and there are times where it’s spot on and more times where it’s blatantly wrong. I like to use it as a rubber duck to help me think and work through problems. But I’ve learned that whatever the output is requires as much scrutiny as a good code review. I fear there’s a lot of copy and pasting of wrong answers out there. The good news is that for now they will need real engineers to come in and clean up the mess.

aabhay 3y ago
Spreadsheets actually did put many accountants and “computers” (the term for people that tallied and computed numbers, ironically a fairly menial job) out of business. And it’s usually the case that disruptive technology’s benefits are not evenly distributed. In any case, the unfortunate truth is that AI as it exists today is EXPLICITLY designed to replace people. That’s a far cry from technologies such as the telephone (which by the way put thousands of Morse code telegraph operators out of business)
- bdw5204 3y ago
  It is especially sad that VC money is currently being spent on developing AI to eliminate good jobs rather than on developing robots to eliminate bad jobs.
  - pama 3y ago
    The plan has always been to build the robots together with the better AI. Robots ended up being much harder than early technologists imagined for a myriad different reasons. It turned out that AI is easier or at least that is the hope.
    - grogenaut 3y ago

cornel_io 3y ago

Two years ago we were quite a ways out from having LLMs that could competently respond to commands without getting into garbage loops and repeating random nonsense over and over. Now nobody even talks about the Turing test anymore because it's so clearly been blown past. I wouldn't be so sure it will be very long before solving big, hard, and complex problems is within reach...

jamaicahest 3y ago

> LLMs in particular can be a very fast, surprisingly decent (but, as you mention, very fallible) replacement for Stack Overflow I think that sentence nails it. For the people who consider "searching stackoverflow and copy/pasting" as programming, LLMs will replace your job, sure. But software development is so much more, critical thinking, analysing, gathering requirements, testing ideas and figuring out which to reject, and more.