February 7, 2023 at 3:27 am
You should have just asked it "How would Jeff Moden do it" - LOL
Ok... I took that challenge. It knows who I am and who I'm asking about and so I have to give it 10 points on the flattery scale but I'm also seriously aggravated that it insists that I'd use the same lame method that doesn't actually work as what it posted before.
I wasn't going to offer to help train the damned thing but, if it's going to use my name in such vain, I may have to. 😀
--Jeff Moden
Change is inevitable... Change for the better is not.
February 7, 2023 at 3:32 am
BWAAAAA-HAAAA-HAAAA-HAAAA!!! I don't recall Itzik ever using the word TALLY in any of his code. At least I'm not the only one that it's mis-quoting.
Ladies and Gentlemen, be afraid... be VERY afraid. 😀 And, for goodness sake, remember what the letter "A" stands for in "AI". It does NOT stand for "Actual". And, for it to come across so bloody cocksure that Jeff Moden or Itzik Ben-Gan would do it that way is horrifying, not because either of us are involved but because people will believe the damned thing on matters much more important.
--Jeff Moden
Change is inevitable... Change for the better is not.
February 7, 2023 at 7:15 pm
BWAAAAA-HAAAA-HAAAA-HAAAA!!! I don't recall Itzik ever using the word TALLY in any of his code. At least I'm not the only one that it's mis-quoting.
Ladies and Gentlemen, be afraid... be VERY afraid. 😀 And, for goodness sake, remember what the letter "A" stands for in "AI". It does NOT stand for "Actual". And, for it to come across so bloody cocksure that Jeff Moden or Itzik Ben-Gan would do it that way is horrifying, not because either of us are involved but because people will believe the damned thing on matters much more important.
They have thumbs Up/Down buttons by each answer that you can click and leave feedback, which they supposedly look at in the aim of improving the system.
February 7, 2023 at 7:59 pm
They have thumbs Up/Down buttons by each answer that you can click and leave feedback, which they supposedly look at in the aim of improving the system.
There are three issues with that... either it will "self train" based on how it's been programmed to self-train by humans that are the original cause of the problem or those humans will try to train it and still get it wrong. The third issue is that if we help it, it will likely be overridden by the legions of "experts" that think they know better. Go look at the number of people that still recommend incremental recursive CTEs as a solution for things that need a numeric sequence.
--Jeff Moden
Change is inevitable... Change for the better is not.
February 7, 2023 at 9:01 pm
Jonathan AC Roberts wrote:They have thumbs Up/Down buttons by each answer that you can click and leave feedback, which they supposedly look at in the aim of improving the system.
There are three issues with that... either it will "self train" based on how it's been programmed to self-train by humans that are the original cause of the problem or those humans will try to train it and still get it wrong. The third issue is that if we help it, it will likely be overridden by the legions of "experts" that think they know better. Go look at the number of people that still recommend incremental recursive CTEs as a solution for things that need a numeric sequence.
I think it was initially trained by giving it petabytes of text taken from the internet, books and other sources which it used to perform machine learning with supervised learning as well as reinforcement learning with Microsoft on their Azure supercomputing infrastructure for several months. I think it works by generating one word or token at a time based on the statistical likelihood of that word occurring as the next word from its analysis of it reading all those documents.
It's very good at writing English but large language models are well known for writing plausible sentences that are not correct. It sounds very confident in its answers but they might be totally wrong. But it is really surprising how good it is.
February 7, 2023 at 11:42 pm
To me, it's really not surprising at how really bad it is. And to give it an authoritative "expert" tone on these code questions is one of the worst ideas I've ever come across. To me, the company has violated it's own usage agreement by making it sound so authoritative. Argue that if you want but that's my opinion.
As I said on another thread, I've never seen an "automatic bullshit grinder" before. 😀
--Jeff Moden
Change is inevitable... Change for the better is not.
February 7, 2023 at 11:50 pm
Jeff Moden wrote:As I said on another thread, I've never seen an "automatic bullshit grinder" before. 😀
Thats because you’ve not met me 😳
BWAAA-HAAA-HAAA!!! You're on my bucket list, Mr. Burrows!
--Jeff Moden
Change is inevitable... Change for the better is not.
February 7, 2023 at 11:54 pm
David Burrows wrote:Jeff Moden wrote:As I said on another thread, I've never seen an "automatic bullshit grinder" before. 😀
Thats because you’ve not met me 😳
BWAAA-HAAA-HAAA!!! You're on my bucket list, Mr. Burrows!
Ditto my ‘ol friend 😀👍
Far away is close at hand in the images of elsewhere.
Anon.
Viewing 9 posts - 31 through 38 (of 38 total)
You must be logged in to reply to this topic. Login to reply