Logan Zoellner


What will GPT-4 be incapable of?

Play Go better than AlphaGo Zero.  AlphaGo Zero was trained using millions of games.  Even if GPT-4 is trained on all of the internet, there simply isn't enough training data for it to have comparable effectiveness.


Yeah, I definitely think we're very early in the transition.  I would still say it's extremely likely (>90%) even given no new "breakthroughs".  

The real-life commercial uses of AI+robotics are still pretty limited at this point.  Off the top of my head I can only think of Roomba, Tesla, Kiva and those security robots in malls.

Anecdotally, from the people I talk do deep learning + any application in science seems to yield  immediate low-hanging fruit (one recent example being protein folding).  I think the limiting factor right now is the number of deep learning + robotics experts is extremely small.  It's also the case that a robot has to be very cheap to compete with an employee making minimum wage (even in developed countries).  If there were 10000x as many deep learning experts and everyone in the world was earning $30/hour I think we would see robots taking over many more jobs than we do presently.

I also think it's likely that better AI + more compute will dramatically accelerate this transition.  Maybe there will be some threshold at which this transition will become more obviously inevitable than it is today.  


Perhaps"when will TAI be developed?" is something that can only be answered retrospectively.  By way of analogy, it now seems obvious to us that the invention of the steam engine (1698) and flying shuttle (1733) marked the beginning of a major change in how humans worked, but it wasn't until the 1800's that those changes began to appear in the labor market.

(Pseudo) Mathematical Realism Bad?

Initially the beings are pure minds existing in a empty universe, so there's no risk of dying or killing yourself, but plenty of driving yourself mad.  If they want a body, they have to imagine it into existence like anything else.  They reproduce by imagining other beings into existence.  I'm not really sure where the first one came from or how it learned anything, but at this point they have a thriving society and a culture for training new minds how to exist in harmony with the others.  One of the chief concerns of the beings is maintaining the norms of this culture with the worst possible punishment being ostracism for people who don't play by the rules.

Objects imagined into existence follow the laws of physics you imagine along with them, so you could have ice that melts or a perpetual motion machine if you want that instead.

It's also possible to create "planes" with more restrictive rules (sort of like spinning up a VM in a computer).

(Pseudo) Mathematical Realism Bad?

Changing the title to "pseudo mathematical realism bad?"

Libertarianism, Neoliberalism and Medicare for All?

There's a lot to get into here, maybe I will start a separate post about "ideal tax policy".  


I think the "ideal" reference case for non-distortionary tax policy is one with zero taxes in which all public services are provided for by a magic genie somehow.

Libertarianism, Neoliberalism and Medicare for All?

Yep.  This definitely not how it's done in the "real world".

In the "seat belts" example, this would involve replacing a law mandating seat-belts what a (presumably high) tax on selling vehicles without seatbelts set to equal the economic/social benefits of seat belts.

I think as a matter of pragmatism, there are cases where an outright ban is more/less reasonable than trying to determine the appropriate tax.  For example, I don't think anyone thinks that the "social  cost" of dumping nuclear waste into a river is something we actually want to contemplate.

Libertarianism, Neoliberalism and Medicare for All?

I think there is a good argument for a general principle in most cases of not subsidizing bad actors for stopping causing harm.

A carbon tax refunded in the from of a UBI is economically equivalent to a "low carbon subsidy" in which each citizen is paid for the amount of carbon they consume below the defined threshold.  In one case we are "penalizing bad behavior" in the other we are "subsidizing people for avoiding bad behavior".  

I agree that for the sake of optics we should "tax bad things" and "subsidize good ones" but from an economic point of view this is irrelevant.


But "avoiding introducing economic distortions" is, to me, not an obvious goal for a tax policy

Given two tax systems which produce the same amount of "income" for the government, we should prefer the one which leads to higher welfare overall.  This is why, for example, raising all of a government's income from tariffs is bad, because it uneconomically disadvantages imports leading to a less efficient economy overall.

A poll tax distorts in favor of the rich, a flat tax or VAT doesn't account for the declining marginal value of money to individuals (regressive in utility if not dollars)

I was suggesting a flat tax on income or consumption (VAT).  These should be identical over a lifetime.  A poll tax would be bad for obvious reasons.

I think that we should solve for "regressiveness" by doing a UBI, not by messing with the tax code.

WRT "declining marginal utility of money", I think this is over hyped.  Rich people don't consume dramatically more than those in the middle class.  To the extent that they spend their "excess" income on charity or investment, the marginal utility of those dollars is possibly higher than giving the same money to a already well-off middle class family.

I would much rather have a UBI in place than a minimum wage, if that choice were in front of me.

I think we agree here

Propinquity Cities So Far

I picked an extreme example of over regulation as a caricature, not to prove the general case.  But needless to say California has also rejected well-reasoned proposals with an ability to make a real impact.

Propinquity Cities So Far

Don't basically all cities control density pretty tightly? I know that a lot of density restriction is just nimbies defending housing scarcity, but it can't all be that, can it?


I think this is where the whole post goes off the rails.

In the real world there are massive economic inefficiencies created by government restrictions on density.  Suggesting that we can fix these with a more complex government system is like suggesting we can solve the "wolves eat sheep" problem with bigger wolves.

Load More