I am doing the reverse, and trying to predict the last year that LLMs use NVIDIA GPUs. It's just an accident of history that video game cards are useful for LLMs, and there is absolutely nothing that NVIDIA is doing from a design standpoint that the big hyperscalers can't do on their own, cutting NVIDIA out, and doing a better job of it as they know their own unique needs. The only advantage NVIDIA has is supply chain relationships and it takes time to establish those, but once that's done, we'll see all the big companies rolling their own silicon and no longer relying on NVIDIA.
You are underestimating just how many people that are out there that want free long distances calls lol. I worked at a phone company and this was a never ending persistent security issue. There are lots of tricky ways to get someone to pay for your long distance call. If the pay phone was free then the local provider would be on the hook for those calls.
Just block long distance calls right? If it was that simple it would not be a persistent issue.
I two am wondering exactly what form slamming the gates shut in our face will take. Closing the first hit is free train And opening the doors to pay me, $#%&
I two am wondering exactly what form slamming the gates shut in our face will take.
"You will rent only the best PCs, eat only the tastiest bugs, and live in the 15-minute City of Tomorrow (also known as New Kowloon). And you will like it. Or else."
Sorry to be the wet blanket. However research on monkeys/apes has for the most part proven that their intelligence is at a dead end and never can progress past what is basically around human 2yo level.
That really depends how you measure and define intelligence and does a disservice to them.
Toddlers for example dont tend to have gang wars for territories and certainly couldnt do battle outcome predictions from a glance at a group across thick canopy and the sounds of branches and hollering.
> However research on monkeys/apes has for the most part proven that their intelligence is at a dead end and never can progress past what is basically around human 2yo level.
How do you prove something like that with animals that can come up with a strategy, form battle plans, execute them, etc.
Even a 4 year old has less strategic vision than what is required to wage a prolonged war over years.
This is not the objective of this book, but The Language Puzzle discusses why primates have never exhibited any verbal language skill as we recognize it past the capability of a infant/toddler, even the best achieving examples we have of primates show they cannot manipulate language as well a child of four or five, and some of those studies with humans raising primates in their homes have some particularly unscientific bits, it also discusses why the vocal abilities of all other primates is lesser than humans and the language centers in brain that we think we know about in humans is also much lesser or not used to the same extent in other primates measured using MRI/anatomy studies, the progression of brain and vocal capabilities of homo sapiens progenitors to develop language via paleontology that shows the divergence from other primates, and many experiments with wild and captive primates of all types to demonstrate some language skill but nothing past very simple meaning for one sound, that might not be common to a geographically separate group, and not always the same meaning for the same group, and the inability of primates to use gestures without lots of prompting for communication. The highest form of communication I remember is one study that shows that orangutangs might be able to communicate a meaning of "in the future" via example warnings about snakes to young ones but you can read about that yourself, it seemed kind of speculative, too. Off top of my head it's a comprehensive overview of primate language research and evolution of physiology of brain/vocal abilities/hands. I don't agree with all the conclusions at end of the book (prior to this everything is based on what we have evidence for so there's a bit of speculation towards the end) but it's fun to think about.
Humans today perhaps. People tend to underestimate our abilities in nature because we’ve evolved to be able to shape it. In reality humans had generationally transmitted oral knowledge of food, plus are the only animals that can transform food at will, including from “toxic” to consumable.
The real question to me here is not the computer. Its why is there such a segment of the population that is so willing to listen to a machine? It it upbringing, societal, circumstance, mental health, genetic?
I know the Milgram obedience to authority experiments but a computer is not really an authority figure.
I think of it as they are additive biased. ie "dont think about the pink elephant ". Not only does this not help llms avoid pink elphants instead it guarantees that pink elephant information is now being considered in its inference when it was not before.
I fear thinking about problem solving in this manner to make llms work is damaging to critical thinking skills.
ehhh... You are getting really pedantic here. Your argument is basically well voting doesn't matter so we should have a dictatorship instead of democracy. Your vote does matter in the aggregate, its not supposed to be a 1:1 outcome. Its an aggregate process. You don't measure the speed of each tire on a car you just measure the car speed overall.
I guess if your position really is votes don't count unless its the one to tip the scale no one would be debating you because its irrelevant. Maybe lead with that.
> Your argument is basically well voting doesn't matter so we should have a dictatorship instead of democracy
I am actually not saying that at all, which is what makes this a paradox. I think democracy is important and that getting an accurate determination of the will of the people is important... but it still doesn't make sense, from a pure game theory perspective, to vote in an election.
It isn't REALLY a paradox, because both things are true - voting is a good way to get a representative sample, and a single vote isn't going to change that sample very much.
In many ways, this is just basic statistics, and we experience this every single election night - elections are called WELL in advance of every single vote being counted, because we already know with statistics what the result will be. Now, there are a few examples of this going wrong, but those are mostly times where the rush to call an election makes people call it when the statistics say there is still a reasonable chance for a comeback.
Honestly, it if was up to me, elections should be determined by a random sample. Randomly select N citizens from the country to be the voters that year, and use that result. If N is sized correctly, you will get the same result you would with an actual election, and we don't have to have everyone waste their time voting.
This would never happen, of course, and I honestly can kind of understand why; the results aren't the only thing that is important, feeling connected to the process and that your voice matters also is important. It is a bit of a agreed upon delusion, though.
reply