Yes, I read that on Reddit yesterday…
Yeah… that’s the usual AI yaddayadda for saying: “I have no clue, I don’t give sh*t, just don’t ask, …” blahblah
1 Like
Although I appreciate the idea of a future AI informing customers by stating, “Thank you, but this will never work,” “Thank you, but you haven’t thought this through,” or “Thank you, this will cost you more than you expect” after they dictated their requirements … 
But from Austin we will get one which says after Compiler run failed:
- I thought in the beginning it should work
- maybe we could correct it in docs
- there are bugs and there are WORKAROUNDS
I’d just be happier if the AI said
Are you sure you want to do that ?
Yes, could also be a good Idea. But for this you only need an IDE which gets the errors in Code mostly before even compiling. Something like this:
recognizes shit before it happens and is not even AI
is that on IntelliJ?? what plugin do you to do that??
That is Netbeans, in IntelliJ it looks so:
no Plugin needed for that, standard behavior of the IDE
I saw that story this morning and my first reaction was 
REALLY ?
Asking something that we already KNOW LIES wether it wrote those papers ?
DUH !
One of our researchers did a demo of ChatGPT with a forensics question on some emails. It was completely wrong until you asked the right questions. I liken it to how people figure out how to use Google or other search engines for the first time - there is a certain ‘way’ of asking the question that will return better results. I don’t expect AI to be any different.
Indeed
Reading that story makes it VERY clear that asking the right question the right way is crucial
I know someone who was truamatized as a sophmore in college … she did a paper where her prof accused her of plagarism based admittedly on a hunch – “there’s something I can’t put my finger on but I know you must have plagarized because the paper is too sophisticated.”
And this was at a prestigious private university.
Just because you’re a prof doesn’t mean you are a clear (or honest) thinker. What the real problem was, was that her then-boyfriend had been in his class the year before, he was a brilliant shooting star and another of the prof’s complaints was that he was comparing her to her super-genius BF. I mean I can’t think of too many ways his reasoning process could be more wrong. She appealed, armed with all her sources, and he wouldn’t even look at them.
The world has never lacked for assholes.
1 Like
sadly
As the Bullshit video series noted (poorly summarized)
it takes very little energy TO bullshit and way more energy to combat the bullshit
Remember: when a humlan ask a question, sometims the other human answers may be “creative” (at bes) to completely wrong !
So, what can do a so-called “AI” ?
1 Like
Also known as “the Xojo feedback way” … 
1 Like
Sort of like our new AI masters!
The key to getting useful output from these chatbots is to understand how to construct a query. You can get usefully low (down to zero) error rates if you know how to do that. You have to break it down into steps for the bot, basically. From what I’ve read, once you get good at that, you are on your way to building autonomous agents, but you then have to contend with the cost of using the API, speed of responses, etc.
It’s a great time to play with the tech if you are so disposed, but I think serious, substantive work is a few months away yet. Progress is rapid, and promising. But right now it’s early days.
My take is that this is a breakthrough tech that has great transformative power for both good and ill but it’s not going to replace devs so much as change the nature of the work over time – and the level of abstraction we typically function at. The open question is whether this will be pervasive or more limited. One can make credible arguments either way. A lot of emergent properties of this tech may not pan out … then again, they may.
Let me say it so the AI text components are sometimes really good it depends what your asking the AI. That AI will program in the near food the near future complete complex software I doubt it. I also don’t believe that in the foreseeable future AI will be the better programmer. What I do believe is that I will be able to program simple programs and routines which users are able to program today. And even if I will be able to program complex software it will also be so the somebody has to control and test the solutions an somebody has to do the graphical development for UI. Today everybody is speaking about ai and nobody wants really to try out what it can do today. So today we have no ai which is able to do that what people expecting set it can do. For now a I can be a short a little helper even in professional programming but it is not a professional programmer and it will not be one.
1 Like