Agreed that AI (in general) is not a dead end, but any particular implementation of AI might be. OpenAI et al are betting billions that their approach will turn out to be the best one, but it really is a bet; there's not guarantee that tomorrow the next DeepSeek won't come out with a better algorithm that obsoletes all their investments.

Apple can always buy out whichever company they decide has what they want. It'll be pricey, but Apple has plenty of cash on hand.

Why will it be "winner takes most"? AI isn't like the Internet where there are network-effects that make first-mover status a huge advantage -- e.g. if I could write a better Facebook than Facebook today, it still wouldn't get used by anyone, since Facebook's advantage comes from its huge user base and my new platform wouldn't have one.

With AI, OTOH, anything the first-movers do, Apple can (eventually) copy and improve upon, a strategy they have used successfully many times in the past. Stepping back and letting others figure out what the works and doesn't work, on their own dime, seems like a good approach. Why burn money on what might be a dead-end, when others are happy to burn their own money for you?

Apple isn't doing nothing -- it's continuing to do the things that it has always done, like selling iPhones and computers and streaming services. Those things have always been profit centers for Apple, and they continue to be.

The other thing that it's doing correctly at this point is not losing its head and betting the farm on AI. Other companies would be wise to follow Apple's example.

Right... I worked for The Beer Store, the brewer-owned private company which distributes beer across the Province of Ontario. Our Premier (roughly equivalent to a State Governor) made a campaign promise of "A buck a beer!".

So, a new empty can cost roughly $0.20 at the time. The law in Ontario is that shelf prices include tax and deposit. So, the can is $0.30 - twenty cents for the can itself, plus another dime for the deposit to make sure the used can comes back for recycling.

Now, on top of that, you have to make a food-grade beverage, pay your excise tax to the federal government, and then there have to be profits for the manufacturer and the distributor/retailer (that would be Brewers Distributing Limited dba. The Beer Store).

Customers would come to me and - with that "I know more than you even though I haven't held a job in 16 years" expertise - tell me that we were going to be carrying "buck a beer" because they voted for Doug Ford (who also cut their welfare increases).

"When do you get it? It's gotta be soon!"

"The first shipment arrives February 31st, so mark your calendar!"

I must have used that line 500 times. Only one person realized that there's no February 31st. To his credit, he had to come back to the store to tell me. LOL

Exactly ONE brewer made the Buck A Beer - Cool Brewing of Etobicoke, in Doug Ford's riding. We were lucky if we got a single case (24 beers) a month. Promise fulfilled... Right.

Anyway... The Beer Store's shelf tags were printed at the distribution center and sent to stores with truckloads of beer and empties in and out. Of course, you always had too many tags you didn't need, and were always short of the shelf tags that you did need. If a tag was outdated and wrong, you have to - ethically if not legally - honour the price. And, of course, if a tag was damaged or lost, there was no tag for that product. All of this hearkened back to The Beer Store's roots as Brewer's Retail where everything was behind a counter and we had a selection wall. In a newer self-service store like mine, this did not work.

Electronic shelf tags were implemented. It was amazing. Snap the tag into place on the shelf. Scan the tag. Scan the product. Press a button. The scan gun would beep and a moment later, the tag would update with the item description and price.

Price changes? Automatically updated on all tags.

Now, something about selling addictive substances: Sometimes someone decides that the item's price is what they have, not what the shelf tag says. And they will argue with you until the cows come home. You get jaded to it.

"That will be $2.25 for the can of Pabst Blue Ribbon 5.9."

"The tag says $1.95 so you have to give it to me for that. You forgot to update the sticker."

"No sir, I assure you that it doesn't. They're not stickers, they're electronic and tied to the POS."

"It says $1.95."

"Sir, if the shelf tag says $1.95 for Pabst Blue Ribbon 5.9, I will give you a full case of it. On the house."

For a moment, they're elated. And then they realize that I'm coming out from behind the counter to call their bluff. In front of the lineup of impatient customers during the daily 10:01AM opening rush. Catcalls. Whistles. Jeers.

Walk over with the dude... shelf tag says $2.25 for PBR 5.9. Now, at this point, I'm annoyed, and I'm not going to short my till $0.30 for him. Or suggest to him an alternative beer that is $1.95 a can. If he'd just passed me all his change and come up a little short, I would have covered it. Personally, out of my pocket, if I didn't have a few nickels and dimes perpetually floating around my cash. I've spent way too much time on both sides of the counter at The Beer Store, so I have plenty of empathy - just don't be an asshole.

Anyway... Dollar stores are dealing with customers who are on the very bottom economic rung, whether from addiction or for some other bad life event. Now, sometimes these people are a nickel away from being able to afford a can of beer - or a jar of baby food. I have seen split tender three ways for a $2 item - $0.50 from returning 5 empty cans, $0.97 by scraping a prepaid Mastercard from last Christmas to the last cent, and then $0.55 from under the sofa cushions or wherever. Unexpected price changes can drastically upset plans these people have made to get a few supplies with their very last dollar.

"I can get a box of Kraft Dinner at Dollarama for $0.50, and two cans of cat food at A Buck Or Two with the other $0.50..." I've seen it, and I've personally lived it.

The shelf tags, especially at a dollar/discount/alcohol/cannabis store of any sort, must be accurate. As an experienced retail manager, electronic shelf tags are simply essential.

You can sell the boss on implementing them with the operational savings, the labour of having to change stickers with every price change. Electronic shelf tags will pay for themselves in very short order.

I'll find out in mid January, lol - it's en route on the Ever Acme, with a transfer at Rotterdam. ;) But given our high local prices, it's the same cost to me of like 60kg of local filament, so so long as the odds of it being good are better than 1 in 8, I come out ahead, and I like those odds ;)

That said, I have no reason to think that it won't be. Yasin isn't a well known brand, but a lot of other brands (for example Hatchbox) often use white-label Yasin as their own. And everything I've seen about their op looks quite professional.

I'll need to test it (its been a while) - I'm using an Intel Arc A310 on my rig where I encode. I've got an AMD 9070 XT in my gaming machine but I haven't really tested encoding on that one.

Yeah, it's not even worth considering for something like 15-20kg. A full pallet in this case is 464kg :) Years worth of supply for me.

The current "AI" is a predictive engine.

And *you* are a predictive engine as well; prediction is where the error metric for learning comes from. (I removed the word "search" from both because neither work by "search". Neither you nor LLMs are databases)

It looks at something and analyzes what it thinks the result should be.

And that's not AI why?

AI is, and has always been, the field of tasks that are traditionally hard for computers but easy for humans. There is no question that these are a massive leap forward in AI, as it has always been defined.

It is absolutely crazy that we are all very very soon going to lose access to electricity

Calm down. Total AI power consumption (all forms of AL, both training and inference) for 2025 will be in the ballpark of 50-60TWh. Video gaming consumes about 350TWh/year, and growing. The world consumes ~25000 TWh/yr in electricity. And electricity is only 1/5th of global energy consumption.

AI datacentres are certainly a big deal to the local grid where they're located - in the same way that any major industry is a big deal where it's located. But "big at a local scale" is not the same thing as "big at a global scale." Just across the fjord from me there's an aluminum smelter that uses half a gigawatt of power. Such is industry.

That "ruler study" was ancient. It's mentioned in peer review at least as early as 2018, and might be even older.

Believe it or not, people in the field are familiar with these sorts of things that you just read about.

Most of these new AI tools have gained their new levels of performance by incorporating Transformers in some form or another, in part or in whole. Transformers is the backend of LLMs.

Even in cases where Transformers isn't used these days, often it's imitated. For example, the top leaderboards in vision models are a mix of ViTs (Vision Transformers) and hybrids (CNN + transformers), but there are still some "pure CNNs" that are high up. But the best performing "pure CNNs" these days use techniques modeled after what Transformers is doing, e.g. filtering data with an equivalent of attention and the like.

The simple fact is that what enabled LLMs is enabling most of this other stuff too.

Jevons Intensifies

AV1 seems like a good codec - I'm always happy to use it though, even hardware assisted encoding is still a bit slow for me. If I'm reencoding videos for my own use I usually will encode to x265 instead, which has a good balance between file size/quality and compression speed.

I've done my first test of buying a whole pallet of filament straight from a Chinese manufacturer. It's a risk - it could be all junk - but if it's usable, the price advantage is insane. Like $3/kg for PETG at the factory gate (like $5/kg after sea freight and our 24% VAT). Versus local stores which sell for like $30/kg.

I'd love to see someone try to 3d print with a filament that melts at 162K. Where do you even buy xenon filament? ;)