The interview shows the CEO is kind of a jerk. He probably shouldn't be put in situations where communication is a requirement, like public interviews that are intended to help achieve an aggressive goal.

It's like he didn't understand he was on air during the conversation, despite the host clearly calling out that there was an audience listening.

The stark response from eBay is certainly understandable, having seen the interview.

Do you really mean that if your git repo were corrupted, restoring a snapshot of the repo from backups wouldn't work? If that's true, then it sounds like your backup system is broken. The hashes after restoring ought to be identical to what they were before the backup.

If git used the files' iNode numbers for its hashes, then I could understand how a filesystem-based backup/restore might not really work; you'd have to backup at the block level instead. But git doesn't use the iNode numbers.

git isn't magical. It only knows files. It doesn't know if you moved the repo, copied the the repo, or restored the repo from a ten year old backup. I have moved git repos around plenty of times, `cp -a`ed directories with repos, tared and un-tared directories that contain repos, and the copies have always Just Worked without any hash mismatches.

mkdir ~/test. cd ~/test. git init, touch test.txt, git add test.txt and git commit. cp -a ~/test ~/test2. cd ~/test2 and check out the backup repo. The backup is valid. Then simulate a disaster with rm -rf ~/test. Then recover from the disaster with cp -a ~/test2 ~/test and you've just restored a repo from filesystem-level backup. The resulting repo works perfectly and its hashes aren't off. git has no idea you deleted and restored under its nose. Try it yourself.

What am I missing? I'm not surprised to be called idiotic, and the shoe often fits. But I'm surprised to be called that over this.

My router's hardware's parts were made in China. Its software was made as a worldwide effort but the team seems to be officially based in the Netherlands. And I'm not asking my government's permission for updating either one. Trumptards and their micromanaging far-left centralized-economic-planners can go fuck themselves. Keep your damn dirty ape hands off my computers, comrade.

Agreed that the Mach-E is a terrible name. But how did they screw up such a guaranteed out-of-the-park home run with an electric Mustang? I mean the whole image of the Mustang is a sporty performance vehicle for the young and stupidly lead-footed. Mustangs are classically known for acceleration and EVs are wickedly good at that. I mean, if Ford were to create a 1965-styled electric Mustang, I shudder to think how many boomers would buy them. They were the dream car of an entire generation.

Ford, are you listening?

They want to wind down sales and not have to invest more money into EVs. The Lightning had been a serious contender for my next vehicle, once it switched to NACS, but now it is just abandonware.

The program is still deterministic - the output is determined *entirely* and deterministically by the input. (Where the input is the set of the prompt, the sequence of numbers returned by the calls for random(), and the LLM data model itself.)

Your "mistake", if we want to call it that, is treating the random() function as an innate quality of the LLM. It isn't it is simply part of the input.

Provide the the system with the same model, the same prompt, and the same sequence of numbers, and you WILL get the same answer, regardless of how complex the question is, or who asks it.

Intel is improving for sure, as is AMD and Micron. Nvidia though is still just a rockstar on all levels. Happy to hold and aquire a "laggard" like Nvidia...

So drop it down in scale: if you get a home equity line of credit does that mean you should pay an asset tax on the value of your home? Sure, it is just for billionaires. Just like the Alternative Minimum Tax.

Something needs to be done with income inequality but I'm not sure this type of strategy will work as intended.

It doesn't matter if it is new or existing; if you need 200A of continuous load you have to install a 600A electrical service... and never load it over 200A. All the Span panel can do is work within your power envelope to leverage that excess capacity... but you can't avoid the 30-40% cap on sustained load vs service capacity.

Is there room for consciousness to manifest in a brain?

What if you write a bunch of random noise of the same file size to weights.bin ... what happens then?

You absolutely can though. There is nothing stopping you from seeding the run with a single LLM, or even substituting the function definition for random() with:

random() { // determined by fair dice roll
        return 5;
}

We can trivially and easily do this.

And further, it seems you are now suggesting that substituting the above random function for this one:

random() { //
    input = ask-user-for-fair-dice-roll();
    return input;
}

and now you sit there rolling dice and inputing the results, and the computer program gains consciousness?

really?

The difference, of course, is that we currently DO actually know EXACTLY how an LLM works. We can snapshot the model and seed the random number generator to make it generate exactly the same output from exactly the same input every single time. We can pause it, set breakpoints, inspect and dump data structures.

It IS simply a program running on a CPU, and using RAM.

Is it possible that's all humans are in the end? Sure its possible, I can't prove otherwise. But we are not remotely in a position to assert that its the case.

You invoke philosophy which is entirely appropriate. There are fairy tales for example of artists painting things so realistic that they come to life. And it poses an interesting question here: is there is a difference between a simulation and a real thing? Can a simulation of life, be "alive"? Or must it forever remain a simulation.

And a related, and perhaps ultimately simpler question is can a *turing machine simulation of life* be "alive".

A lovely illustration of the question:
https://xkcd.com/505/

Can what you and I perceive as our lives, the universe around us, and everything REALLY be underpinned by some guy in a desert pushing pebbles around in a big desert somewhere?

Can the arrangement of stones in a desert, and some guy updating moving them aorund, in some pattern he interprets as representing the information that describes our universe actually "BE" our universe?

Is is the pattern of rocks is JUST a pattern of rocks. Is the guy moving them around JUST moving them around. Is the interpretation of the pattern as a representation of the state of a universe, just that, a representation?

Or you truly think there is a galaxy with a planet with people on it having a conversation on slashdot,'frozen in time' waiting for some guy to move the rocks into the next pattern and that somehow results in the experience we are sharing right now?

Or put more succinctly - can an abstract representation of a thing be the thing? be it bits in a DRAM module memory or pebbles arranged in the sand? can it be the thing it represents? Can the painting of a zebra if its done skilfully enough be a zebra?

As a matter of fact, yes. Specifically, to the Greenland region.

Although a golden bridge of grossly outsized proportions and festooned with tacky ornamentation would be much preferred.

The parent poster acknowledges this, they are saying the randomization is *introduced artificially*.

The same as any dice rolling app. All you have to do is seed the pseudorandom number generator the same for each run, and it will roll the same dice, in the same order, every time.

Likewise, if it wants to spit out the next word/phrase and 2 of them have 33% probability, and two have 17% ...

Then if you seed the random number generator with the same seed for every instance / run, you'll get the same output from the same input on the same model.

The system is entirely determininistic. The same as any other software, from the ghosts in pacman to the bots in quake arena, to a chess engine. We introduce "randomness" to make it more enjoyable, but its pseudorandomness, that we artificially insert. We could just as easily seed the random number generator the same way every time, and then it would do the exact same thing every time. None of these are actually thinking and making decisions.