I was purposefully giving it the best possible reading to show that even under those generous assumptions that it falls orders of magnitude short from the sorts of revenue it needs to be generating to keep working.

The USD is still the world's reserve currency. Almost all international trade is demarcated in USD. This means a weak USD is a bit bad for the US on balance; but it's devastating for the rest of the world. Tbills are a fundamental part of most central banks' portfolios. A crash in Tbill returns is, again, bad for the US but devastating for the world. The US is also a major oil producer. When oil is above ~$120/barrel, shale oil and gas is suddenly economically viable again. Considering oil is traded in USD and Trump's war has already taken out about a third of the oil production in the region, and the imperialist bullshit Trump has pulled with Venezuela, the US is set to be the world's largest oil exporter. All this is to say that, "When the US sneezes, the world catches cold," is still true. Except the US is going to go into a recession, possibly a depression. The rest of the world was just starting to fully recover from 2008 in 2020 when covid hit. I'm honestly not sure if it can take it if the US goes into recession again.

Even being as large of an economy as it is, the US still punches well outside its weight economically.

It's not a ban if you can still use it. It's not a ban if you can still sell user created data for training new models. Until they actually ban GPTs and enforce said ban, I'm not donating to Wikipedia again.

And for the record, I regularly have been donating whenever Wikipedia has a funding drive for the previous eight years.

You can create a stable plasma with $100 worth of electronics. It's really not that difficult. You could do it as a High School physics lab.

OpenAI lost $11,500,000,000 last year. There are 52 weeks in a year. $100,000,000 over 6 weeks is $866,666,666.67 per year; that's less than a billion. Even if that was just 20% of possible revenue, that's still just $4,333,333,333.33 per year. Meaning OpenAI would still be losing $7,166,666,666.67 per year. And that's assuming it doesn't lose users because of the ads.

Due to other market conditions, I'm predicting that the hyperscalers are all going to crash sometime in late 2028 and take the rest of the US market (and thus the world market) down with it. It will make 2008 look like minor market correction. Not a recession but a full blown depression.

It will not. Part of its generation function involves injecting white noise into the system for randomization.

The word is being used in a technical argot. It's meaning is different than the colloquial understanding. In this context, it means that there are a finite number of definitive states that the procedures must result in. The system must drive users to one of those end states. GPT systems, by their stochastic nature sometimes hallucinate and run off on wild tangents. Deviating even slightly from the prescribed outcomes results in a failure state for the system. If the GPT has a 5-10% hallucination rate, then that's a 5-10% failure rate of the software.

Meanwhile, humans are intelligent. Having a human in the system who understands what the purpose of the system is and the need to drive users to one of the prescribed endstates.

This reminds me of a possibly apocryphal story about a compsci security class. First day of class, the professor walks in, tells the whole class, "Congratulations, you've just failed this class." He then shows them him entering 0s into the database software. "This course is 16 weeks long. Whether you pass or fail is up to you."

I agree totally. The economics of this device are just fucked, which anyone who ever looked into carbon capture could have told you.

A more interesting use for this technology is to hook it up to a high carbon source. Something like a metal refinery or a cement plant that generates large, concentrated amounts of CO2. It then results in a fuel you can use either in surface transport or in regeneratively running the industrial source itself, significantly shrinking its carbon footprint. Ironically, if you were to hook an industrial scale version of this up to a coal-fired powerplant, you could significantly reduce the carbon footprint of extant power generation while building out green alternatives.

The only price is of the device, which is listed as $15k-$20k. At that price, given gas prices, that's a payback timeline of almost 40 years.

An average box fan will move around one thousand cubic feet of air a minute. Converting that to metric, that's about thirty-seven cubic meters of air a minute. A cubic meter is one thousand liters; thus an average box fan moves 37,000L/min of air, roughly. You can very easily move enough air to run a simple hydroxide CO2 capture device. It wouldn't be terribly efficient and it would also be a power hog. Given the size of the machine in the pictures in the article, the numbers they give are reasonable. Even the price seems on par with a back-of-the-envelope estimate. This device doesn't make economic sense but it does make mechanical and chemical sense.

This thing generates one gallon of gasoline per day. As this reads like a press release, I assume that would be under ideal circumstances. For the sake of argument, let's just roll with that. Assuming a price per gallon from a gas station of $2.50 USD, at a price of $20,000 this device has a payback timeline just short of 22 years. I somehow doubt this machine will run for that length of time without encountering the need for maintenance and repair. This would extend the payback timeline to closer to 30 years. The average lifespan of a car is about 20 years. Meaning before you reach break-even on this machine, you'll probably be replacing your car. If you replace your gasoline car with an electric car, the payback on a solar array is just 5 years.

Havana syndrome doesn't exist. It's a psychosomatic mass hysteria. Havana syndrome hasn't been "conclusively proven" false, only because it's virtually impossible to prove a negative. Several scientific bodies looked into it and found it to likely be psychosomatic. Intelligence agencies and the State department looked into it several and determined that it couldn't possibly just be in their heads and so issued multiple wildly differing theories with little to no evidence in their favor.

This is a very good idea. It isn't a short term solution to anything but it would be a long term net benefit to the economy. It would also go a way towards the de-financialization of the economy and controlling consumer debt loads. Which is why none of Trump's handlers, party members, or financiers will allow it to go through.

It's so weird to me that Trump the person has a habit of throwing out good ideas I assume he's heard somewhere, immediately walking them back, then doing the opposite. You'd think the media would have figured out this pattern. Instead, they Charlie Brown for it everytime he pulls a Lucy.

Adams was a moron who tried to treat his cancer with dewormer and apple cider vinegar. Much like Steve Jobs, if he had been half as intelligent as he thought he was, he would have listened to subject matter experts and treated his cancer with chemotherapy, radiation, and/or surgery as his oncologist recommended. Instead, he died as a testament to the fact that money rots your brain and your soul.