How do you quantify "incentive"? Is a landlord really looking at 5% lower property value and deciding it's not worth investing? Is this even true in aggregate?
It is a major paradigm in economics that if you change this by X% it will change something else by Y% and to estimate that ratio. It may be that people don’t really think that way: economic growth seems to be continuous and exponential in character whereas economic dislocations are discontinuous in character.
I think of how I was absolutely shocked when a Big Mac meal was $10 during the pandemic (I think it cost about $2.50 the first time I bought it) and didn’t think I was going to buy 4% less of it but rather I skipped the fries.
Well, there's 2 ways to become a landlord: to buy a house or to build a house. I was focused only on the second way.
The cost of the wood and the labor needed to build the house is unaffected by the rent control, so if cost remains the same, but the reward (or "revenue") from building the house decrease from $100K to $95K, then fewer houses will get built.
So they're losing out on a few dozen new units, but it is made up for by renters across the board having to pay less rent and thus having more money to put into the economy? Seems like a good trade to me?
Because then the units fall into disrepair (because they no longer make sense to maintain) leading to less supply leading to lower availability of units leading to higher cost of housing.
See what happened in NYC regarding the consolidation of housing units.
It means only 2% of the harmful rays (UVA) are getting through the shirt or alternatively the skin under the shirt can spend 50 times as long in the sun as it could without any protection.
Yudkowksy gave up on trying to make a god-in-a-box to stop other gods-in-boxes in 2015. Since then his approach to stopping the gods-in-boxes has been to lobby governments.
It is easier for me to play speed chess with smooth animation of each move rather than when a piece instantly teleports from origin to destination, but I have reason to believe that I'm unusually intolerant to frequent activation of my orienting response.
Either that or the only reason they've been releasing the models under permissive licenses is that that the only way they have get any attention in a market dominated by American companies.
(Also, they don't need to make a profit because their system does not prioritize profit potential when making investment decisions: it prioritizes alignment with directives out of Beijing, which include keeping up with the West in strategic technologies.)
It remains possible that Altman and Amodei are sincere: they might believe that AI is dangerous like nuclear weapons, but there's no way to stop its development (especially since it would need to be stopped globally, not just in the US and countries the US can influence) and they consider themselves to be more likely to do a good job of it than their competitors.
"LLMs can never become out-of-control superintelligence" might be relevant if there weren't many hundreds of researchers working (at OpenAI, Anthropic and elsewhere) on AI designs not based on the transformer (LLM) architecture.
People are working on lots of things all the time, so far, nothing has approached the efficacy of the transformer architecture.
LLMs didn't emerge by chance, they are the culmination of decades of research intersecting with brute force engineering rigor in a perfect storm of innovation. You're not just going to stumble into an alternative approach by dumping loads of cash into research.
Most of those researchers would disagree with your statement that the next major breakthrough is probably decades in the future.
Many of the insights accumulated in the decades up to now will probably prove useful for creating non-LLM AI, and the researchers can use LLM AI to speed up their research into non-LLM AI.
>Isn't every nuclear bomb an existential threat to all human life?
No, of course not. It's a threat to people within 15 miles of the explosion plus people who are outdoors and turn to look at the bright light in the sky.
And there's never been enough nukes to kill all human life. That statement is based on a despicable calculation in which it is assumed that people would assemble packed shoulder-to-shoulder in circles of just the right size and there are no structures or land masses to deflect the blast and no clouds or fog to absorb the intense light.
reply