Let's flip the argument around then. Assume for a second their AI division is perfectly altruistic, and then somehow writes up a perfect political plan to deal with the consequences of AI.

The people and goverments of world would have no way to trust their good intentions, no way to check if their plan is good, and no intention of carrying it out.
So after having gone through all that effort, the politicians will still decide what happens.

Why would it be the job of the AI companies to deal with the political changes of improved technology?

We didn't ask the same for the inventors of the printing press, the steam engine, or combine harvester.
You let the inventors bring out disruptive technology, then it's up to people with both political power and responsibility deal with the changes.

You know, politicians. (or back in the day, monarchs.)

I know it's hard to keep politics and economics separate, but this is pretty purely political news and should not be on the economics subreddit, even if i do approve of it.

It was only after the release of half life that Gabe Newell got his way and valve released the smash hit and worlds greatest videogame, Ricochet. /s

Think the exec made the right call here.

Was able to generate two free images with the new model earlier today in europe.

Was fun to see this be the first model that could handle my silly benchmark prompt properly.

Any developing technology will have teething problems.

Once we have good self driving cars at affordable prices (my guess, 2 years), i give it another 5 years before every new car sold has full self driving capability, then you need ~20 years before all the non self driving cars are junked before they can make it mandatory.

Small, rich and dense cities will have mandatory self driving earlier, similar to emission zones now.

No need to retaliate with car tariffs, could retaliate with bourbon tariffs, or software tariffs, or agriculural.

Whatever is least damaging for them and most damaging for the US. they are probably crunching the numbers to find out what that is.

When most people interact with AI, they are doing it with integrated tools like these that are often months behind the SOTA.

IMO the first freely available AI that was worth using in daily life were deepseek r1 and gemini 2.0 flash thinking, which was only 2 months ago. (this is crazy to think about)

Most people have been interacting with much older and smaller models.

Give it 12 months and i'm betting the opinions on AI will have shifted immensely.

Programmers/ software developers are probably the only people who could make prices like this work.

If AI doubles the productivity of your 100k a year salary programmer spending 5k per month is not off the table.

But at o1-pro pricing, asking a simple question to help pickout a good healthcare plan, which used about 5k tokens for me on gemini, would cost about 3$.

3$ for one question.
That's a price level where it's better to either do the work yourself or find a paid human expert to answer your questions.

Programmers/ software developers are probably the only people who could make prices like this work.

If AI doubles the productivity of your 100k a year salary programmer spending 5k per month is not off the table.

But at o1-pro pricing, asking a simple question to help pickout a good healthcare plan, which used about 5k tokens for me on gemini, would cost about 3$.

3$ for one question.
That's a price level where it's better to either do the work yourself or find a paid human expert to answer your questions.

Wasn't aware they were also doing rate limited api and not just the consumer facing service.

Yeah if you can't pay for more usage and instead face rate limited api they would need to split their benchmark across many accounts, which is possible but annoying.

No api, so instead of being able to do the benchmarks automatically someone has to feed them into the prompt box 1 by 1.

Consistency is definitely getting better, but not quite there yet, the headset appears and disappears, and his outfit isn't the same in each shot.

Definitely approaching the point where it might get past the point of "AI slop" and we enter the era of AI art though. especially with proper human guidance and editing.

The API for this will definitely not be free, remains to be seen how cost effective it is, though i suspect it will still be somewhat reasonable.

Probably advised by his lawyers to get out as soon as he realized it was legitimate to minimize any chance of being charged with espionage. He would have probably much preferred to stay in the chat as long as possible to get more info otherwise.

R1 is still using the old v3, they are probably working hard to prepare R2 for release as we speak though, and this is sure looking promising now.

I don't think he would be painting cryptic versions of it based off someone telling him.

That only seems to make sense if it's memories that have leaked through the severance procedure, similar to reintegration.

That's weird since innie Irving has not seen the hallway before.
That's a mystery that can now be resolved since we know someone can have multiple innies.
My guess would be a previous version of innie Irving went through the export hall and "retired", and outie Irving is now seeing his "dying" moments in his dream.

Continuing to choke mark wouldn't have worked because goat lady would have shot him, leaving mark alive.

He could have actually surrendered at that point, but instead he knocked the gun away and kept fighting until he was well and truly beaten. If he had won the fight it would have gone poorly for mark and goat lady.

Except it's not just russian oligarchs that lose trust, it will be all the worlds sketchy billionaires, europe and us included, that will wonder if they are next when they seriously misbehave.

Even separately from any allegations of direct foreign bribery, this would erode trust in the belgian financial institutions that hold much of the russian money, which could cost the belgians a lot due to loss of trading fees.

That doesn't mean they shouldn't do it anyway, but it does explain the resistance from the people that prefer cash over security.

Gold is the one metal i don't think the US will ever have a shortage of for use in electronics because it's been hoarded for it's value.

If there was ever a real shortage and the US tapped its gold reserves for industrial usage they would have enough to supply the entire worlds industrial usage of gold for 25 years.

Can't really think of any other metal for which such a huge stockpile exists.

House pricing is cost of (correctly zoned) land + construction costs.

If the sum of those prices rise above what people can afford, developers will not eat the cost and sell at a loss, they will simply stop building. see: 2008.

Yeah, house prices rising does very little for you as a normal home owner. nice payday if you own dozens or thousands of houses though.

Makes you wonder if the current administration has ties to the real estate industry /s.

Think of something like speedrunners running on a certain version of the game.

They will often avoid new patches for 2 reasons. The first is a level playing field for comparing runs.

But the second which is valid for people that just play for fun, is that you can't reach true mastery in the game if you change it every week. there's a pure joy to be had from playing something enough to get really good at it, even if the game itself is flawed.