Sorry, I hit the wrong Reply link... this wasn't in response to you.

In short, because audiences hate it. High framerates destroy the suspension of disbelief and go too far into the Uncanny Valley.

The telling problem in this case is who's asking the question, because it's not the people with the technical expertise. It's the managers who "heard that AI will revolutionise software development".

It reminds me of a Dilbert strip where PHB strings together some Blockchain buzzwords.
Alice: "Did you recently read an article?"
PHB: "Some of it."

One of the very earliest pieces of advice I was given by a heavier user of LLM coding tools than me was: if it doesn't give you what you want at the first or second shot, start over or give up. The whole back-and-forth of trying to coax it toward the solution is a fool's errand. The model is either able to give you what you want almost instantly, or it's not capable, period. I find that this advice still holds very true.

However, with very actively developed libraries and frameworks, I often run into the problem of LLM tools generating code for deprecated interfaces, or mixing the APIs of different versions.

It won't compile until someone publishes a malware package under that name.

No, he's the kind of person who thinks that a market valuation of $400M for an upstart pet food distributor might be a tad over the top.

It was pretty obvious even before the dot-com bubble burst who provided some genuine value and could find a viable business model, and who was just riding the hype wave.

Same as now. Beyond fancy fuzzy search (which is valuable and probably viable, sure) I have yet to see such reasonable looking use cases for LLMs and GenAI.

The midterms are almost a year away still, right? I have strong doubts that the AI bubble won't have started truly popping by then. Time isn't on their side.

Also, how can SBF still be funding anything?

Well, those following LLM recipes deserve everything they will eat.

LLMs can help as a fuzzy search, to find good, human sources. But anyone not seeing the problem in letting them come up with a recipe shows an utter lack of understanding about how these things work. It's beyond me why anyone would want to follow a recipe that hasn't been developed and repeatedly cooked by an entity with hands, a kitchen, tastebuds and a digestive system.

Stuff like this is almost always just about conservatives getting triggered by some buzzwords without understanding, nor wanting to understand, what's actually behind it. As it is in this case.

What this admittedly weirdly worded statement was about was a collaboration initiative with an Indigenous college to research impacts on coastline Indigenous settlements. In other words, it's about trying to get out of the Ivory tower and talk and work with the regular folks who actually know the reality on the ground, are affected most by the decisions, yet often not listened to by the "establishment".

You know, the thing right-wing populists love pretending they're for when trying to win votes, but always work against when on office.

It's just another example of right-wing voters cheering on the stuff that *they themselves* will suffer from the most. Just claim something is "woke" and they'll grab their pitchforks, willingly ignorant of the fact that the work they're so upset about was conducted to try and maybe save their live some day. California and New York will be fine. It's the rural areas that are losing disaster protection, medical services, health inusrance, access to impartial media, etc.

And who is going to be interested in those randomly generated "famous" people? Who is going to pay to see movies with them in it? Who wants to hear music made by something else than a human person? I'd guess it's not very many, and for those who do, it's the novelty factor that intrigues them, which will be over at some point.

Replacing humans with algorithms in the arts is just the next wet dream of predatory tech bros. How much those are in touch with reality and humans is symbolised nicely by Zuck's conviction that you can just replace human relationships with statistical token generation and everything will be hunky-dory.

Just look at how much of the movie, music, literature, etc. business is about interviews, fandoms, meet and greets, concerts, festivals, award shows, signed merch, readings, etc. etc. to get an idea of how unimportant people really think the humans behind the art they enjoy are.

I'm regularly in contact with students here, mostly CompSci, since I represent my employer at job fairs. Over the past few years I noticed a sharp increase in students worrying about whether they'll find a job after graduating. And the anxiety probably come from concrete events. The big US tech companies with offices here are in a headcount reduction phase, after many years of expansion (and promises that they'll continue to expand). There are local options, of course, but the employment dynamics in the US tech sector are definitely having an effect here. I don't blame young people for more likely going to trade school now. The feeling in the air is that there are too many high-tech workers right now, not too few.

But what should they have chosen instead? Check the state of Nvidia drivers on Linux and Mac. AMD was absolutely the only sensible option.

Sure the result is unfortunate, but I respect AMD for taking the risk of trying to do the right thing, and for helping (once more) expose how idiotic the HDMI Forum is and why the HDMI standard absolutely needs to die and be replaced yesterday.

The million cables are an annoyance for sure, and things like eARC are a great convenience. But this stuff is usually set up once and not touched for many years. Also, there are good programmable remotes to get everything back under one control. I think it's not much of a dealbreaker.

I was planning on a new home cinema setup a year ago. I ended up dropping the project, but had already reached the point in my research where the general shittiness of HDMI made me give in to the realisation that I'd have to configure video and audio completely seperately. In the forums I researched in, it seemed to have been the consensus that it's the better choice anyway, and people would end up regretting the convenience solutions with HDMI and co.

We have no such thing as artificial intelligence. We have LLMs. We have next token predictors, we have statistical image generators and fuzzy search. While occasionally helpful, it has not even a remote relation to intelligence, and sci-fi AI is still as sci-fi as ever.

And who's gatekeeping here? Creatives produce content with GenAI and it's the public who lets them know that they hate it. I guess you could call the audience the ultimate gatekeepers, but that makes it sound unnecessarily conspiratorial. People read books, listen to music or watch movies because they're interested in what other people have to say. Without human creative work behind this, there's no point to it. It's vapid, bland, incoherent, boring and hackneyed. I stop reading any text that I suspect to be GenAI immediately because it's a complete waste of my time. That's not gatekeeping, just my conclusion to what has some inherent worth and what doesn't.