zogwarg

I wouldn't be so confident in replacing junior devs with "AI":

1. Even if it did work without wasting time, it's unsustainable since junior devs need to acquire these skills, senior devs aren't born from the void, and will eventually graduate/retire.
2. A junior dev willing to engage their brain, would still iterate through to the correct implementation for cheaper (and potentially faster), than senior devs needing spend time reviewing bullshit implementations, and at arcane attempts of unreliable "AI"-prompting.

It's copy-pasting from stack-overflow all over again. The main consequence I see for LLM based coding assistants, is a new source of potential flaws to watch out for when doing code reviews.

Pessimistically,
I feel TechBroism is a brand of positivism that will never die, more than one of its brethren is already trying to cast themselves as would-be alchemist promising gold from lead through the arcane uses of AI: "You just need the right prompt."

Honestly this could be an improvement compared to what is currently in use by the current french tax collection agency.

The DGFiP uses an up until recently closed-source custom language called "M", which does not have the friendliest/most readable syntax, and that the guys at INRIA (French National Institute for Research in Digital Science and Technology, the same lab that seems to have spat out CatalaLang) had to reverse engineer a modern compiler when open sourcing the tax calculation software was newly mandated.

Witness this horrid glory, sadly only in French: chap-1.m

Could also be intended for other horrid COBOL output cases:

For example, the compiler can generate Javascript for web applications, SAS for economic models and COBOL for legacy environments

Trying to approach and make visible the relationship with the laws as written, so that it can potentially be reviewed by non domain-experts doesn't appear to me to be the worst possible goal out there. (They seem to be trying interleaved markdown format), the bigger/broader claims in the about/readme sections might just be required bells and whistles for proper grant funding/thesis presentation.

I have developer co-workers who play the role very convincingly ^^, especially for the LLM hype, and I remember more than one tech conference attending evangelizer (to be fair sometimes the tools or practices are actually good).

What’s the focus of UX despair these days? I haven’t touched frontend design or implementation for a long while now.

Am I the only one surprised they managed to enter the server rack room in the first place?

More pessimistically i'd call it the "hype train", a bubble promises an eventual pop. What usually happens is people on masse moving the newer and shinier model, apparently afflicted with amnesia about their previous buzz words.

I'd say it's the cancerous influence of VC funding, the grift gets the dough.

Through the magic of self-funded blockchain crowd combobulator (famously very low on greenhouse emissions) yet vitally out-funded through the commitment of big oil money (also famously very low on greenhouse emissions), any regulation or actual government action to address climate change is unneeded!

Truly a pair, paragons of ~~sustainable~~ web3/libertarian virtue!

Quite the roller coaster of the author swinging wildly on both ends of the sneer line.

Not sure about the people who “are or think they are intelligent” being more susceptible to the con.

It feels like something one wishes to be true for karmic/poetic reasons rather than something that actually IS true.

I think good marks for the LLM con are more generally doubtful of the value of human intelligence/labour/education and/or tech positivist rather than true believers in their own intelligence.