I use AI for rather complex tasks. It's impressive. It can make a bunch of non-trivial changes to several files, and have the code compile without warnings. But I need to iterate a few times so that the code looks like what I want.
That being said, I also lose time pretty regularly. There's a learning curve, and the tool would be much more useful if it was faster. It takes a few minutes to make changes, and there may be several iterations.
It sounds like the guys in this article should not have trusted AI to go fully open loop on their customer support system. That should be well understood by all "customers" of AI. You can't trust it to do anything correctly without human feedback/review and human quality control.
This is just an incredible statement. I can't think of another development tool we'd say this about. I'm not saying you're wrong, or that it's wrong to have tools we can't just, just... wow... what a sea change.
If there is such a tool, programmers will be on path of immediate reskilling or lose their jobs very quickly.
Imagine if your calculator app randomly and non-deterministically performed arithmetic incorrectly, and you similarly couldn't get correctness expectations from the developer.
Imagine if any of your communication tools randomly and non-deterministically translated your messages into gibberish...
I think we'd all throw away such tools, but we are expected to accept it if it's an "AI tool?"
They keep telling you that any employee who highlights problems with the tools are just trying to save their job.
Your investors tell you that the toolmakers are already saving money for your competitors.
Now, do you want that second house and white lotus vacation or not?
Making good tools is difficult. Bending perception (“is reality”) is easier and enterprise sales, just like good propaganda, work. The gold rush will leave a lot of bodies behind but the shovelmakers will make a killing.
If you think of AI like a programmer, no we shouldn't throw away such tools because we accept them as imperfect and we still need to review.
Of course there is a review system for a reason, but we frequently use "untrusted" tools in development.
That one guy in a github issue that said "this worked for me"
Because no other dev tool actually generates unique code like AI does. So you treat it like the other components of your team that generates code, the other developers. Do you trust other developers to write good code without mistakes without getting it reviewed by others. Of course not.
Literally yes. Test coverage and QA to catch bugs sure but needing everything manually reviewed by someone else sounds like working in a sweatshop full of intern-level code bootcamp graduates, or if you prefer an absolute dumpster fire of incompetence.
2) No matter what the learning curve, you're using a statistical tool that outputs in probabilities. If that's fine for your workflow/company, go for it. It's just not what a lot of developers are okay with.
Of course it's a spectrum with the AI deniers in one corner and the vibe coders in the other. I personally won't be relying 100% on a tool and letting my own critical thinking atrophy, which seems to be happening, considering recent studies posted here.
Where the AI fails is in doing anything which requires having a model of the world. I'm writing a simulator which involves agents moving through an environment. A small change in agent behaviour may take many steps of the simulator to produce consequential effects, and thinking through how that happens -- or the reverse: reasoning about the possible upstream causes of some emergent macroscopic behaviour -- requires a mental model of the simulation process, and AI absolutely does _not_ have that. It doesn't know that it doesn't have that, and will therefore hallucinate wildly as it grasps at an answer. Sometimes those hallucinations will even hit the mark. But on the whole, if a mental model is required to arrive at the answer, AI wastes more time than it saves.
I wonder if anyone has compared how well the AI auto-generating approach works compared to meta programming approaches (like Lisp macros) meant to address the same kind of issues with repetitive code.
Not op but yes. It sometimes takes a lot of time but I read everything. It still faster than nothing. Also, I ask very precise changes to the AI so it doesn’t generate huge diffs anyway.
Also for new code, TDD works wonders with AI : let it write the unit tests (you still have to be mindful of what you want to implement) and ask it to implement the code that run the tests. Since you talk the probabilistic output, the tool is incredibly good at iterating over things (running and checking tests) and also, unit tests are, in themselves, a pretty perfect prompt.
Opposite experience for me. It reliably fails at more involved tasks so that I don't even try anymore. Smaller tasks that are around a hundred lines maybe take me longer to review that I can just do it myself, even though it's mundane and boring.
The only time I found it useful is if I'm unfamiliar with a language or framework, where I'd have to spend a lot of time looking up how to do stuff, understand class structures etc. Then I just ask the AI and have to slowly step through everything anyways, but at least there's all the classes and methods that are relevant to my goal and I get to learn along the way.
Unless you're thinking of repetitive code I can't imagine the process (I'm not arguing, I'm just curious of what you're flow looks like).
Generally, all the code I write is reviewed by humans, so commits need to be small and easily reviewable. I can't submit something I don't understand myself or I may piss off my colleagues, or it may never get reviewed.
Now if it was a personal project or something with low value, I would probably be more lenient but I think if you use a statically typed language, the type system + unit tests can capture a lot of issues so it may be ok to have local blocks that you don't look in details.
But when I see people using these AI tools to write JavaScript of Python code wholesale from scratch, that's a huge question mark for me. Because how?? How are you sure that this thing works? How are you sure when you update it won't break? Indeed the answer seems to be "We don't know why it works, we can't tell you under which conditions it will break, we can't give you any performance guarantees because we didn't test or design for those, we can't give you any security guarantees because we don't know what security is and why that's important."
People forgot we're out here trying to do software engineering, not software generation. Eternal September is upon us.
2) I find the tool analogy helpful but it has limits. Yes, it’s a stochastic tool, but in that sense it’s more like another mind, not a tool. And this mind is neither junior nor senior, but rather a savant.
Then it's not a useful tool, and I will decline to waste time on it.
The LLM too. You can get a pretty big improvement by telling the LLM to "iterate 4 times on whichever code I want you to generate, but only show me the final iteration, and then continue as expected".
I personally just inject the request for 4 iterations into the system prompt.
The vibe coding guy said to forget the code exists and give in to vibes, letting the AI 'take care' of things. Review and rework sounds more like 'work' and less like 'vibe'.
/s
