What is fascinating about the new #AI#LLM revolution is that a storm is coming, the experts are telling us, we can see it, and it will be fascinating to see how industry reacts.
In short, #programming as a profession is going to largely die. I hear numbers like "in ten years" being bandied about, though I'm skeptical of the timeframe.
Developers are the 21st century version of the well-paid #Luddite textile workers, except we have years of advance warning,.
TIL that * is a valid left hand side/ local part of an email address according to RFC5322. I knew for a long while that a quoted at sign "@" also is a valid local part of an email address.
Also TIL that Microsoft365 doesn't allow you to use *@<domain> as an email address or alias, and GMail rejects "@"@<domain> as an email address.
So if you want to forcefully defederate your email from Google and Microsoft, you know what to do.
It'll likely break 87% of all other email services too though.
1966: Ein Beamter verweigert vor Gericht seine Aussage, weil ein #Kreuz im Saal hängt. Recht und Religion seien Dinge, die der Staat nicht verquicken dürfe. Zitat der kath Presse: "Der Terror, den ein Häuflein von Atheisten gegen die christ Bevölkerung ausübt, wird zum Skandal."
While testing a #Perl script on a couple of #Linux servers, a strange error message popped up on one specific server:
Can't locate strict.pm: Permission denied
Usually the "Can't locate xyz.pm" error message points to a missing Perl module. But strict.pm comes bundled with Perl (perl-libs) itself and is not part of a separate module. So why would the script fail to locate strict.pm?
@ck And if you want to see where your perl looks for modules, run perl -le 'print for @INC'. (This includes paths from the PERL5LIB and PERLLIB environment variables.)
Haskellers - n00b question: I want to share an array of data between a background task that refreshes this array periodically and a REST api that requires read only access to it. Should I create a global TVar and share it between these two parts of the application? What would be a better way? Or is this an acceptable simple solution? Thank you. #haskell#stilllearning
Updated my #Macbook Pro and it again replaced my backticks with the useless § symbol. I've also lost my tilde. Really, really ticked off that #Apple keeps doing this.
Doesn't help that I'm using a French keyboard remapped to a US one and system settings doesn't seem to let me actually set the damned things.
@ovid Mine seems to be in some sort of "split brain" mode where on the external keyboard, the backtick symbol is on the key left of "1" (above tab), but the built-in keyboard has it on the lower left, between left shift and "z".
In recent months I’ve noted a certain “fuck you, change my diaper” mentality that crops up again and again with regard to “AI”
What typifies this is a whirlwind of goalpost juggling, a sort of desperate, insatiable righteousness gasping to justify an existing feeling:
fear
The more I see it, the more I suspect a whole subset of technical practitioners are afraid of losing what they built their identities on: being special computer boys
@danilo I find your point of view interesting, but it doesn't seem compelling to me. I am a special computer boy, but I don't feel threatened by "AI" that tells me "010 is an invalid octal number" because "octal digits are 0-7" (that's a real example from last week). And that's not an isolated case: I regularly find nonsense (sometimes obvious, sometimes very subtle) in LLM-generated code and its accompanying explanation. To me, this seems like a fundamental issue with trying to use a chatbot to write code. Some of it is going to work, some of it is going to fail in obvious ways, and some of it is going to be subtly but fundamentally flawed. If you're not already a domain/programming expert, you might not be able to tell the difference, so the LLM ends up adding negative value to the process.
I may be wrong. Maybe some specialized models will improve to the point that they no longer produce nonsense, but there's no money in that. It's much easier for them to produce code with bugs that are increasingly harder to spot, all with plausible-looking comments and an authoritative "explanation".
Let me slightly reword one of your earlier comments:
"The particular copium spew that [cryptocurrencies/NFTs] cannot possibly be useful, and that those who feel they are helpful have been deceived or deluded? That’s about GATEKEEPING. That’s about feeling [banking] democratize and fearing that it comes at your expense" — I don't believe it when it's about cryptocurrencies, and I don't believe it (yet) when it comes to LLMs.
What thriller and/or mystery TV series / movies / books have people engaging in reasoning that actually makes sense?
Almost every example I can think of reminds me a bit of how little kids reason, e.g., in Jack Reacher, they're constantly trying to show how smart Reacher is with little kid reasoning, like when he "smartly" concludes that the bad guys didn't find what they were looking for because some place was trashed. Sure, or maybe they did and were checking if there was more, were mad, etc.
@saagar Wasn't Encyclopedia Brown the one with "contradictions" like "magicians pull stuff out of their sleeves, therefore all magicians must wear long sleeves, but X wore short sleeves!"