Eric Schmidt deleted Stanford interview • Hacker News Board

Re: Eric Schmidt deleted Stanford interview

#101

Post by wordofx » Mon, Aug 19, 2024, 5:06 AM UTC

inferiorhuman wrote: post #41

tialaramex wrote: post #25

Earlier quoted context omitted.

> you're not going to let people work from home and only come in one day a week Huh, I guess I dreamed the first start-up I worked for (a couple of decades ago) where indeed I only came in one day a week. Yes of course you "work like hell". We had a nasty leak bug and I set things up so that day or night if the leak was detected my stereo would go maximum volume and play "Straight Outta Compton". How does commuting c…

If I'm sat in a car There are other ways to commute. Commuting via train gave me a chance to go for a walk through parts of the city I'd otherwise not spend a lot of time in. Being stuck in suburban hell currently, working 100% from home is a nightmare scenario for me. Asshole neighbors. Constant noise (far worse than e.g. Oakland). Shit infrastructure (electric, telecom, whatever). And, in my experience, collaborati…

There’s no point in arguing with people on HN. The same people downvoting you are the same people who refuse to use AI because it’s “not perfect”. They would rather work from home and be made redundant and unemployable than be productive and valuable.

Re: Eric Schmidt deleted Stanford interview

#102

Post by genocidicbunny » Mon, Aug 19, 2024, 5:09 AM UTC

inferiorhuman wrote: post #41

tialaramex wrote: post #25

Earlier quoted context omitted.

> you're not going to let people work from home and only come in one day a week Huh, I guess I dreamed the first start-up I worked for (a couple of decades ago) where indeed I only came in one day a week. Yes of course you "work like hell". We had a nasty leak bug and I set things up so that day or night if the leak was detected my stereo would go maximum volume and play "Straight Outta Compton". How does commuting c…

If I'm sat in a car There are other ways to commute. Commuting via train gave me a chance to go for a walk through parts of the city I'd otherwise not spend a lot of time in. Being stuck in suburban hell currently, working 100% from home is a nightmare scenario for me. Asshole neighbors. Constant noise (far worse than e.g. Oakland). Shit infrastructure (electric, telecom, whatever). And, in my experience, collaborati…

> There are other ways to commute

Really? Pray tell what other methods I would have to commute? The closest bus, train or subway service is about the same distance as driving to work with a car. Parking at work is also free, but I have to pay for parking at the bus stop, or the train stop, or the subway stop. Oh and there's no way to walk there because I would have to walk along the shoulder of a 4-lane highway where the speed limit is 65mph.

So, how do I commute without a car?

Re: Eric Schmidt deleted Stanford interview

#103

Post by keiferski » Mon, Aug 19, 2024, 5:09 AM UTC

flawn wrote: post #100

keiferski wrote: post #91

It will be interesting to see if Schmidt’s comments about the future of AI being mostly a China-US game will come true, or if less powerful models will ultimately be more useful than single centralized powerful ones. The analogy here might be between nuclear weapons and drones; the former are controlled entirely by a small number of countries, but the latter probably have more of a direct role to play in the future o…

Check Leopold Aschenbrenner's work out - he introduced this to the public through AI Safety https://news.ycombinator.com/item?id=40576466

That’s a pretty big PDF - can you point me to the part where he talks about what I’m talking about?

Re: Eric Schmidt deleted Stanford interview

#104

Post by habosa » Mon, Aug 19, 2024, 5:11 AM UTC

Eric ran Google while the Android team often worked 7 days a week in office (famously: “bacon Sundays”) and still managed to blow every advantage they had over iOS by making repeatedly terrible strategic decisions.

But yeah, he’d have won AI by keeping butts in seats…

Re: Eric Schmidt deleted Stanford interview

#105

Post by genocidicbunny » Mon, Aug 19, 2024, 5:14 AM UTC

wordofx wrote: post #101

inferiorhuman wrote: post #41

Earlier quoted context omitted.

If I'm sat in a car There are other ways to commute. Commuting via train gave me a chance to go for a walk through parts of the city I'd otherwise not spend a lot of time in. Being stuck in suburban hell currently, working 100% from home is a nightmare scenario for me. Asshole neighbors. Constant noise (far worse than e.g. Oakland). Shit infrastructure (electric, telecom, whatever). And, in my experience, collaborati…

There’s no point in arguing with people on HN. The same people downvoting you are the same people who refuse to use AI because it’s “not perfect”. They would rather work from home and be made redundant and unemployable than be productive and valuable.

No, the people arguing with the parent are the ones who have actual life experience of living in places where there's no reasonable access to public transit (read: most of the USA) and who aren't naive to think that everywhere is a perfect little urban world where you have usable choices in your method of transport.

I live inside a major metro area. My closest public transit is a 10 minute drive by car, or a several hour walk by foot, and that's a 3x a day bus. If I wanted to actually access useful public transit, I would need to hike 3+ miles over about 1500' of elevation change, on roads with no sidewalks. And I'm in one of the most public transit friendly metro areas in the US.

Re: Eric Schmidt deleted Stanford interview

#106

Post by ChuckMcM » Mon, Aug 19, 2024, 5:21 AM UTC

dlg wrote: post #72

ChuckMcM wrote: post #31

Earlier quoted context omitted.

Eric has a penchant for saying things he thinks before thinking about how they might be heard. He famously said, "There is no privacy, get over it. If you're not doing anything wrong you don't have anything to worry about."

Wasn't that Scott McNealy? (Though, if I recall your bio from previous HN posts, you'd know far better than me.)

Just great, now I'm getting so old I'm mixing up my Sun Executives :-).

You are correct. I also believe Scott said something to the effect that IT was dead although we all know how good of a prediction that was.

Re: Eric Schmidt deleted Stanford interview

#107

Post by jemmyw » Mon, Aug 19, 2024, 5:21 AM UTC

zooq_ai wrote: post #84

nextos wrote: post #24

Earlier quoted context omitted.

Lots of great startups have remote teams so this argument by Schmidt is pretty weak. Google is hybrid, not even remote. Also, I don't think that well motivated and engineers work or produce less while WFH. It could be the opposite. Less interruptions are key, as explained in Peopleware . But bad managers love controlling people in open plan offices. Regardless, I found it surprising Schmidt didn't talk about other st…

can you name successful remote-only startups? Yes, I know gitlab, others?

Aha! https://www.aha.io

Re: Eric Schmidt deleted Stanford interview

#108

Post by MobiusHorizons » Mon, Aug 19, 2024, 5:29 AM UTC

zooq_ai wrote: post #84

nextos wrote: post #24

Earlier quoted context omitted.

Lots of great startups have remote teams so this argument by Schmidt is pretty weak. Google is hybrid, not even remote. Also, I don't think that well motivated and engineers work or produce less while WFH. It could be the opposite. Less interruptions are key, as explained in Peopleware . But bad managers love controlling people in open plan offices. Regardless, I found it surprising Schmidt didn't talk about other st…

can you name successful remote-only startups? Yes, I know gitlab, others?

I believe tailscale is remote only. Personally remote doesn’t work for me, I feel my motivation and mental health go into a tailspin after a while, but there is no denying some companies have made it work.

Re: Eric Schmidt deleted Stanford interview

#109

Post by wunderlust » Mon, Aug 19, 2024, 5:30 AM UTC

"The country is going to have to learn critical thinking. That may be an impossible challenge for the US." (ts 30:57)

"Impossible" is a bold qualifier. Maybe he was exaggerating - maybe not. Will critical thinking necessarily become more difficult - to learn, teach, exercise - with AI? It's possible it will help people become better thinkers, but I don't think that's a guarantee.

Re: Eric Schmidt deleted Stanford interview

#110

Post by mikewarot » Mon, Aug 19, 2024, 5:41 AM UTC

At about 39 minutes in, he's asked if efforts analogous to seti at home can be used to get around the scaling problems with training the next round of big models.

He gives a strongly NVidia oriented answer that I happen to think is dead wrong. Pushing more and more GPU/Memory bandwidth into more and more expensive packages that are obsolete after a year or two isn't the approach that I think will win in the end.

I think systems which eliminate the memory/compute distinction completely, like FPGA but more optimized for throughput, instead of latency, are the way to go.

Imagine if you had a network of machines, that could each handle one layer of an LLM with no memory transfers, your bottleneck would be just getting the data between layers. GPT 4, for example, is likely a 8 separate columns of 120 layers of of 1024^2 parameter matrix multiplies. Assuming infinitely fast compute, you still have to transfer at least 2KB of parameters between layers for every token. Assuming PCI Express 7, at about 200 Gigabytes/second, that's about 100,000,000 tokens/second across all of the computing fabric.

Flowing 13 trillion tokens through that would take 36 hours/epoch.

Doing all of that in one place is impressive. But if you can farm it out, and have a bunch of CPUs and network connections, you're transferring 4k each way for each token from each workstation. It wouldn't be unreasonable to aggregate all of those flows across the internet without the need for anything super fancy. Even if it took a month/epoch, it could keep going for a very long time.