I think this places too much weight on the words open and free. K3 is very much not free in the conventional sense. It's about 30% cheaper than GPT 5.6 Sol, on openrouter right now. It's true that companies could run it on their own infrastructure but that's also not free, and also true that this price will come down, like the prices for all models.
Thanks, Sam — I agree with you about all this, but the threshold for my argument going is not that open models are free in all respects; it's that that create significant downward pricing pressure on the market. A 30% gap on Open Router, which is going to be doing a lot of the selecting for companies anyway, is exactly that
OpenAI’s externally stated goal is already to “make intelligence too cheap to meter”. This post is well written but greatly underestimates the amount of change that is likely to come.
I'm down with the massive change scenarios; my point is to underscore the high uncertainty and what it means for funding a welfare state to handle those changes
The Constitution requires just compensation for any government taking. The public would be paying market prices for Sanders’s proposed 50% stakes in these companies, which could very well be a bad deal (in addition to the separate arguments for why government ownership of market participants is a bad idea). Should we be forced to buy SpaceX shares at current market prices too? Why on earth should Americans be forced to take investment advice from Bernie Effing Sanders???
Just as we can't be sure where the profits for an AI boom might show up, it also seems like we can't be sure that the traditional means of taxation will be able to capture those profits. The US has an unusually high rate of tax compliance. AI could disrupt that. An AI boom might mean large reductions in people with W2s. It might be much easier for people to keep their income or capital gains private or obscured from the IRS. It may make it much easier to operate out of more tax competitive jurisdictions.
Electricity generators generate electricity not "electrons"! I will die on this hill. It doesn't make you sound smart to use the incorrect word it makes you sound ridiculous.
It may be the case that corporate taxes today are too low, but that discussion should be entirely separate from one around taxing AI returns.
The three bets you referenced are necessary to consider separately because, should they be true, they would mean that the big labs would capture massive monopoly-like value. Without them, more or less the status quo but at a higher gdp.
Supposing the bets are wrong and AI value is distributed as you argued throughout the economy:
Traditional taxation is most effective, but it also adds additional burden to corporations operating under relatively intense competition no different systemically than corporations today. It may be true the rate should be different, but AI doesn’t play into it. We should just figure out the best corporate tax rate to balance revenues and growth, with or without AI.
Should the bets be correct:
The big labs end up being some of the most influential and lucrative monopolies in history and a unique policy approach is warranted.
I think this places too much weight on the words open and free. K3 is very much not free in the conventional sense. It's about 30% cheaper than GPT 5.6 Sol, on openrouter right now. It's true that companies could run it on their own infrastructure but that's also not free, and also true that this price will come down, like the prices for all models.
Thanks, Sam — I agree with you about all this, but the threshold for my argument going is not that open models are free in all respects; it's that that create significant downward pricing pressure on the market. A 30% gap on Open Router, which is going to be doing a lot of the selecting for companies anyway, is exactly that
OpenAI’s externally stated goal is already to “make intelligence too cheap to meter”. This post is well written but greatly underestimates the amount of change that is likely to come.
I'm down with the massive change scenarios; my point is to underscore the high uncertainty and what it means for funding a welfare state to handle those changes
Agree that the govt. should NOT make these bets. Consistent with Sanders' ignorance of how capitalism works.
The Constitution requires just compensation for any government taking. The public would be paying market prices for Sanders’s proposed 50% stakes in these companies, which could very well be a bad deal (in addition to the separate arguments for why government ownership of market participants is a bad idea). Should we be forced to buy SpaceX shares at current market prices too? Why on earth should Americans be forced to take investment advice from Bernie Effing Sanders???
Indeed, it seems likely that a nationalisation in part of US AI groups would have a better than even odds of tax payer buying in at top of market...
Just as we can't be sure where the profits for an AI boom might show up, it also seems like we can't be sure that the traditional means of taxation will be able to capture those profits. The US has an unusually high rate of tax compliance. AI could disrupt that. An AI boom might mean large reductions in people with W2s. It might be much easier for people to keep their income or capital gains private or obscured from the IRS. It may make it much easier to operate out of more tax competitive jurisdictions.
Electricity generators generate electricity not "electrons"! I will die on this hill. It doesn't make you sound smart to use the incorrect word it makes you sound ridiculous.
thanks beau
i’m personally a fan of the poetic license on this, but given that it’s also cliché to say electrons, i think that license can be safely revoked
Hoping for 2/2 for the meat-slicing deli owner when the accrual of gains in the capital class shakes out on this one 🤞🤞
It may be the case that corporate taxes today are too low, but that discussion should be entirely separate from one around taxing AI returns.
The three bets you referenced are necessary to consider separately because, should they be true, they would mean that the big labs would capture massive monopoly-like value. Without them, more or less the status quo but at a higher gdp.
Supposing the bets are wrong and AI value is distributed as you argued throughout the economy:
Traditional taxation is most effective, but it also adds additional burden to corporations operating under relatively intense competition no different systemically than corporations today. It may be true the rate should be different, but AI doesn’t play into it. We should just figure out the best corporate tax rate to balance revenues and growth, with or without AI.
Should the bets be correct:
The big labs end up being some of the most influential and lucrative monopolies in history and a unique policy approach is warranted.
The 50% number coming out like the exact reverse of the Dr. Evil “One Million Dollars”.