Perl Monk, Perl Meditation | |
PerlMonks |
Re: Is ChatGPT worth $42 a monthby haukex (Archbishop) |
on Jan 27, 2023 at 09:40 UTC ( [id://11149921]=note: print w/replies, xml ) | Need Help?? |
I think this is pretty good! Can you define "pretty good"? Though its answer sounds like it could have been written by a human, it is riddled with errors and inaccuracies. And from my understanding of how AI like this works, that's completely understandable - there is no YAPE::Regex::Explain built into it, instead it has simply read a ton of different explanations of various regexes and is only regurgitating those snippets of what it has read that it associates with regexes that look similar to the input, no matter if those snippets actually apply or not. Is ChatGPT worth $42 a month For creative writers who don't care about factual accuracy, it's a great tool; anyone requiring accuracy will be disappointed. Though of course, lots of people will try and use it for science and technology... ((?:(?:[^\n@]+|@[^@]*@)\n?)+) This regular expression is used to match a pattern that contains multiple lines of text. It can match multiple lines of text, but doesn't have to, like this sentence suggests. one or more other characters (also represented by @) This makes no sense, and if it's trying to say that @ represents one or more other characters, that's of course wrong. It should say "one or more characters other than an @ sign" instead. The outermost parentheses ( ) define a capturing group that captures the matched text, this group will be repeated one or more times + . The second part of this sentence is of course wrong - or if I'm feeling very generous, at the very least inaccurate, since it's the contents of the capturing group that'll be repeated. The first non-capturing group (?: ) is used to match any characters except a newline or an at sign. The second non-capturing group (?:@[^@]*@) ... Wrong twice. The ? after the first non-capturing group ... The ? after the second non-capturing group Wrong two more times. This regular expression should be used in conjunction with other tests or validation to ensure that it only matches valid input. It is not a complete solution to validate emails or other texts, just a pattern matching. This underlines my point quite well. The first sentence is of course good advice, but the second sentence, while not wrong, misses the mark: I don't think this regex has anything to do with emails (and calling a regex "a pattern matching" sounds weird). The AI will be good at general creative writing tasks and usually sound like a human, thus it'll fool a lot of people, but the AI doesn't care whether it's accurate or not.
In Section
Meditations
|
|