spunker540

I recently learned about the battle of Midway on one of the recent WWII Netflix docuseries.

It was an early battle following Pearl Harbor, and it was Japan’s entire carrier fleet vs US’ entire carrier fleet, and the grand totals were something like 4 carriers vs 3 carriers.

At the end of the battle they mention that by the end of the war the US was completing multiple carriers per month, and they were just orders of magnitude ahead of every other country when it came to manufacturing power and steel production. Japans only hope was finishing the us Navy early on at Pearl Harbor and midway but once the US ramped up production the US never lost their military lead (arguably they still have it). Meanwhile at the start of WWII the US wasn’t even top 10 militarily. It’s astounding how much changed in those 5 years.

I'm not too convinced about the lobbying argument. I know they spend $2-3M per year on lobbying. but 1) our govt doesnt get anything done anyways, regardless of how much Intuit spends. I believe the null hypothesis is pretty likely here (they could not spend that money and still be the #1 tax company and no real laws get passed to take that away from them). 2) $3M just doesnt sound like a lot to a company as big as Intuit. If that's all it takes to get congress on your side, seems like not that big an obstacle in the first place for any company in any industry.

I am open to more direct evidence to the contrary though.

I think we've all learned to conflate these things from lived experience, not some mis-placed prejudice against ugly UIs.

If “spouting off” includes incriminating yourself then there’s nothing stopping law enforcement from using what you say in the public square as evidence in court. Thus, if IG is a public square, then scraping IG is also fair game for producing suspects.

They’re probably irked by how AI pessimists keep moving the goalposts

The average American watches close to 3 hours of tv per day— I just don’t see what the fuss is about. It seems like you simply don’t approve of how other people choose to spend their time.

Why do you think complex algos, complex feeds, across large search space (all content instead of friends and family) turns out to be more profitable? Are ads suddenly more valuable in that context? I don’t think so. And it’s certainly much more expensive to operate than a simple chronological feed of friends and family.

It comes down to time spent. People spend more time on the apps and see more ads. We can argue over whether that’s a sign people like the app more… but no one is being forced to scroll a feed they don’t enjoy. Interesting that people “are begging for changes” yet using the apps more and more anyways.

If the generative part moves entirely client side, then video calls could become extremely low bandwidth. You could basically stream just the audio of the conversation, and each person could still feel like they were still talking face to face through photo realistic avatars. I don’t how likely this is, but maybe it makes reliability so good that it’s a no brainer. You could probably not even look at your camera if you’re busy, but the other person will still feel like you’re looking at them the whole time.

They may not sing but their popular tracks still have vocals don’t they?

Are researchers actually frustrated to never see it launch, or are they mostly focused on publishing papers?

I thought OpenAI’s unique advantage over many big tech companies is that they’ve somehow figured out how to fast track research into product, or have researchers much more willing to worry about “production”.

I guess I’m just curious what the killer use-cases are for fine tuning. For example it seems like overkill to fine tune a Shakespeare model, because you can just say “write like Shakespeare” and it already knows what you want.

Is there a consensus yet on when you should fine-tune vs when you should use prompt engineering?

It seems not that hard to include in your prompt “here is some examples, please write like this and follow the style set here”. While that may make every completion request more expensive (more tokens), it also seems like fine tuning these models can also be quite expensive.

I’m curious if there are other trade offs besides cost— maybe quality achieved is better with fine tuning? Very interested to see how it all plays out. On the one hand a massive model like gpt4 can probably be prompted to match any style quite well, albeit costly, vs fine tuning a cheap model may get exactly what you want.

It’s a lot easier to proofread something than to write it from scratch.

Also from a computational complexity angle there’s actually a mathematical definition — “NP Complete” which describes problems that are cheap to verify but hard to solve. One example could be a solution to a maze. It’s easy to look at the maze solution and verify it connects start to end without crossing walls, but much more time consuming to actually solve the maze with all the wrong turns needed to produce a solution.

Won’t it find traction in bullshit jobs pretty easily?

Most big co’s make it pretty clear to employees that it’s a pretty big no-no to discuss proprietary work through an unaudited 3rd party (especially a competitor).

Another similar concept I’m aware of is magicians without borders

I don’t think it’s a big deal if your AI spits out copyrighted images. My scanner, camera phone, photoshop etc can also easily spit out copyrighted images. It’s ok if I just print them out for my wall at home, but if I try to sell them it becomes a problem.

So then I guess these new AI models are only problematic if someone tries to sell them (as its essentially selling copyrighted works) but an open source model is probably fair game. Individuals can use it, but if you try to monetize you better not be infringing.

Seems like something like YouTube’s content ID solution for detecting copyrighted works could be a “quick” fix here. You can sell access to your model but not to generate copyrighted works

It is small I agree. It I think it’s like if a chess engine wins against a grandmaster a handful of times, it’s enough data to know that it works.

These newspapers all have Facebook pages (and Twitter accounts) where they voluntarily post and promote their news articles. I’d venture they get a ton of free traffic from social media companies and are far more dependent on social media than the inverse…

The bottom line is people weren’t paying newspapers for their news. They had a geographic moat, and controlled printing and distribution; and when the internet made distribution costs free, their business model tanked, much like the music industry and CDs. But it’s far easier to blame Facebook than admit that the world changed and left them behind.

Lol— I had a friend whose parents took this approach. The parents eventually gave in because they got tired of never being able to reach their son through failed attempts to call friends (like me) in an effort to reach him.

Also big difference between an 8 year old and 15 year old (still not really working age but definitely socializing independently age).

Nobody is denying advertising works. But most of the time ads are trying to win you over vs competitors in the market. Most of it is predicated on the fact that you are in the market for something, and they want you to choose them over someone else. I doubt coca-cola is targeting non-soda drinkers in their advertisements. But they want you to pick coke over Pepsi.

Even if you create an open source drug that cures cancer and give it away for free, you’re gonna need a marketing campaign so that doctors know about it, or that patients know to ask their doctor about it. Not because you’re malicious and trying to manipulate people but because otherwise people who would benefit from your product won’t know it exists.

Can you elaborate on why it's a "dirty word"? I see a ton of potential with generative/procedural game elements, esp given all the recent AI breakthroughs. But I am more of a former gamer than current gamer and am surprised to hear it's a dirty word.

There are certain areas more likely to be vaporized than others though. I know in NYC at least there were some who felt moving to a less populated area (ie rural Midwest) might lower their risk of nuclear death.

Pretty rare for me to see a case with Apple where I feel it’s the other party being overly litigious about trivial naming!

OP’s point was probably that liberals upset about free markets is expected, while conservatives upset about free markets is rather more ironic

I’d be in favor of banning facial recognition as evidence used in trials, but still allowing it to aid in investigations. I.e let it help narrow down suspects when faced with grainy footage, but can’t be used to actually make a conviction.

If social media is to blame for the rise in suicide, why has it only affected American teens when it is a global trend?

I understand at Bloomberg they’ll do an internal “fantasy financial league”, but they rank by sharpe ratio, which adjusts based on the volatility/risk (something like that).

But that’s such a good point about stock competitions generally showing kids that the reckless gambler wins the most - never thought about that.

This is largely irrelevant, but Leonard Cohen’s song “Chelsea Hotel #2” is the primary reason I find the hotel special. Every time I walk past the song pops in my head. I’d be excited to see the hotel returned to some glory, bohemian or not. It’s been under scaffolding for most of a decade it seems.

Jackass hinted at the direction culture was taking with no-names getting famous over short shocking clips, but they also managed to stay relevant and draw huge crowds post-YouTube.

In reality I don’t think it was ever solely about the stunts, but the personalities and the jokes and the pranks. A random YouTuber doing the same stunts won’t compare without humor and personality.