I don’t know anything about ChatGPT but maybe it is possible to use it to generate faster Monthly Reports ? And save time to waddlesplash to work on more interesting features.
These monthly reports are not a simply a sum of the commits done during a period.
It’s a way to say that there’s light in the house and a project proof of life.
They also present progresses to the community in a readable form. It’s important because we users are not all devs or even techs.
Even if ,as users, we expect a monthly report, a job can take a week more to complete, a human being will delay the report so it makes sense.
Some modifications of the code can eventually spread over several periods. In these cases particularly, a human being is able to tell you what’s the plan is and what is expected to come next. I don’t know any kind of tool that can predict future reliably.
So, the report itself has to be done manually, it doesn’t mean that the person doing it can’t use tools to prepare it but it’s not so easy.
it’s a short essay , about progress, it’s a trivial task for a high school kid
Please don’t. In that case I’d rather just look at the git commit log.
I really hope the devs won’t.
Maybe you should learn about how it works before suggesting to use it everywhere.
Can it write activity reports? Yes, sure. Will the reports actually report on what was done in Haiku? No, it will make stuff up. That’s how it works. It’s a “bullshit generator”. A very convincing one, but still.
“Human beeings may not be perfect, but a computer program with language synthesis is hardly the answer to the world’s problems.”
- Deus EX, 2000
I am in full agreement with @pulkomandy above, and also in the post he wrote earlier this month on another thread:
Let me also just say that this is not the place for extended anti-‘AI’ screeds and manifestos, but if it was, I would have already posted one which argued the case on multiple fronts, all the way from the practical (it is wildly inaccurate, and often very subtly so, to the point where you have to think very carefully about its output to notice the problems), to the philosophical (it is clearly less intelligent than most 6-year-olds, even if it sometimes regurgitates the right combination of words to pass university entrance exams) to the humanistic (I don’t want to cede any more of life to the machines than we already have; aren’t writing and thinking and creating good and noble things to do, not merely things we should rush past in an attempt to get somewhere else faster – and where or what, exactly, are those things which we are trying to get to faster, anyway?)
I don’t have an OpenAI account. I hope to never have an OpenAI account, or a Bing Chat one, or have any other interaction with such systems. I don’t even use Apple Siri or Hey Google, for that matter. So, no, even if such systems were capable of writing activity reports accurately and quickly, I still would not use them.
I think AI can at some point in the future partially replace human coders. But I also think it might have a paradox effect: It will perhaps increase the appreciation of human coding - the people and the process.
Two other aspects are there:
- Programming is really fun. Don’t let this fun disappear!
- Do we (the society) really want to let machines take over everything?
(EDIT: I posted this simultaneously to Waddleplash saying this is not the place for anti AI manifestos.)
Well, I said “extended” manifestos, and then wound up writing a mini-manifesto of my own, anyway, or at least a summary of one…
The job of a programmer is not just to turn a specification into code. It is, first and foremost, to carefully analyse the specification, and point out the imprecise or impossible parts in in, and clarify that and design the right software architecture. This is the hard part of the job. Which, so far, ChatGPT is completely unable to do (it will never reply “I don’t understand” or “there is a contradiction in what you say”).
So, will we see improved code-completion systems based on neural networks? Yes, maybe. But that is not the part that takes a lot of time and, especially, effort, when coding something. All the other parts are: testing the code, making sure you understand what you wrote (harder if you didn’t write it yourself line by line, I guess? But that is also the case when you copy code from a documentation, another project, a StackOverflow reply, …).
But, in any case, this will not replace human “coders” in the near future. Because if it did, you would find that the input you have to give to your robot coder has to be so precise that… it’s easier to write it in a properly formalized programming language.
I agree totally. That’s why I said “partially” and why I think appreciation of human coding might increase. Because it becomes more visible what the human brain is doing.
in the span of 10 yrs, we went from no AI that could write code and comprehend text, to that. in evolutionary terms, that around 65m yrs of evolution in a decade, and the current gpt3 version is 3 yrs old.
your view of this is very honestly biased towards humans. there’s nothing wrong with that, but, it’s incorrect.
AI will surpass 90% of human programmers probably in the next iteration, and surpass all human coders in 2 generations. if i was speculating about the future.
we’ve been seeing new AI iterations about every 24-36m.
btw, this is all very worrisome, we don’t actually understand the internals of the AI
My 50ct: No.
Human being can’t stand the comparison anyway. Not because of AI will become “smarter” but because humans are becoming dumb.
The proof is that computers do already better than most humans at human languages. Spell checkers are everywhere and most people are using them. What makes it worst is the fact that these tools are far from perfect. Yet since their apparition the number of human spelling mistakes on simple dictation doubled. Same thing is happening with maths and almost everyone needs a calculator to divide two numbers. Where’s the need to exercise our memory when there are Youtube tutos for whatever that you need to do, Wikipedia for everything that you need to know.
If something is done by computers, humans are simply dropping the ability of doing it. It’s called evolution. We are next to the day, we will become veggies.
Some of us are trying to resist step by step but, you know Resistance is futile…
This frees up more brain capacity to do other things instead. Where’s the problem in that?
Also you can find the same arguments from a 1000 years ago with the invention of books and abacus.
It was irony, of course.
But, I must admit that I have not a lot of faith in human being to use that extra brain power for something useful and even less to share results.
In the office I work we have a software which produces many problems. More than the former one. We have the new one for around two years now and all of us have to use much of our brain to get this software into the right direction, so that is usable. It’s getting better over the time but it is far away from being good. This software “takes care” that we do not forget to use our brains.
I don’t think that we get dump and computers rule the world. Many people are reading books and also manuals or tutorials. Probably more younger people use Youtube and stop reading. The older ones are also able to count more than two numbers, because they have learned it at school. I am an older man and can count, read and think.
It depends on the people themselves if they like to learn their whole live (like me) or to sit on the sofa and watch Netflix.