The passing of Dolly Parton has really had me thinking how much good these unfathomably rich people and corporations could do. It enrages me to think of how many people $106 million could have helped.
Hear hear! I am a nonprofit tech professional, and a monthly call with ~40 similar folks yesterday started with a discussion of Dolly's passing. At first I thought it was just about loving her music and/or personality, but it was also the fact that several folks, at different nonprofits with varied missions, had received significant funding from the many foundations Dolly had set up over the years.
What's really interesting is when you look at how much good they could do without ever actually affecting their lifestyle at all. What actually is $100 million to a trillionaire? What could he do with 1,000,000,000,000 that he can't do with 999,900,000,000?
The problem with this logic is that it equally applies to someone who owns 999,900,000,000.
If someone is willing to spend money above some spendable threshold, then they will never become a trillionaire. Because they will start giving away sooner.
I’m not seeing the problem you’re describing. It has a name: the marginal utility of money. $1,000 is worth a lot to someone who has nothing, less to someone who has a million, practically nothing to someone who has a billion.
So, yes, it applies equally at lower levels, until it doesn’t.
We can walk and chew gum at the same time. The real world doesn't require us to negate the commentary that the ultra wealthy people/corporations could be doing a lot more good than they're currently doing on the basis that the government is spending money irresponsibly/against the greater good.
Are you familiar with the flock camera situation in the States? The American surveillance state has rapidly been getting worse with the help of this technology.
The United States has propped up some pretty horrific, murderous, bloodthirsty regimes over the past century to ensure they maintain a proxy presence across the globe. If supporting or engaging in an "authoritarian" regime is your litmus test, then the United States fails resoundingly.
The unfortunate concequence of "rebellion", is sometimes those regimes go on to be just as bad as the previous regime. You're playing captain hiehnsight here... intent matters in these things.
The fact that the USA only helps right wing "rebellions" is pure luck, right?
Those regimes weren't "as bad as the previous regime" as unexpected result, those resulta were exactly what the US expected.
Guess what you can do in the US and can't do in China? Talk about their respective atrocities. You've missed the point entirely as it pertains to AI if you thought my litmus test was about having the most moral government. It was about that government's ability to censor information.
This is an interesting comparison, but the resulting product of an LLM aren't deterministic in the same way machinery in an assembly line is. You typically expect to get the exact same thing every time as an assembly line worker, and it's usually fairly obvious when something is wrong or off.
So while I agree with your invocation of Marx and the alienation of labor, but I don't know if we really can compare it to the automation of manual labor.
Why is everyone still operating under the assumption the current token costs will remain so heavily subsidized? We could see $200-400/hr in token costs once these companies need to turn a profit
Because then people will just flock to open weight models that are even cheaper and Anthropic/OpenAI will go bankrupt. Then somebody will buy those companies for cents on the dollar and run a profitable token selling business for today's prices without doing any new research. New model releases will slow to a crawl but that's it.
That chart doesn't make the case you think it does. Real median household income rising can be explained by things like more dual earner households (more women working since the 70s), more hours worked, etc. The household income can rise while the wages can theoretically remain flat or even fall.
The more relevant statistic is that median real wages have only grown by about 29% across 40+ years (~0.6% per year)
Since 2000, medical care costs have risen by 121.3%, hospital services by 275%, college tuition and fees by 196%, compared to consumer goods by 86.1%. Things like TVs and electronics went way down in costs while the essentials have absolutely skyrocketed. The cheap stuff drags the average down.
You need a lot more than a single graph to argue against the quality of life going down for Americans.
You're right, I was mixing up the related charts. Still, this makes my case even stronger since personal controls for adding more earners to a household. If real median personal income only rose from ~$28k in 1974 to ~45$k today, that's a 60% increase. Median personal income rising by 0.9% every year over 50 years compared to healthcare rising by 3-4% every year since 2000 is not a gap you can ignore. Necessities grew at nearly 3 to 4x the income rate.
So the case that quality of life is trending downward is still completely valid and shows why you can't just point at a single graph and say "see? line go up therefore quality of life fine"
>Still, this makes my case even stronger since personal controls for adding more earners to a household. If real median personal income only rose from ~$28k in 1974 to ~45$k today, that's a 60% increase.
Labor force participation rate for both males and females has basically been flat for the past decade, so the recent discontent about "costs are all increasing while wages are stagnant or worse" is still unsupported. Moreover the 1970s was never an era of stay at home moms. Female labor force participation rate was already 45%, against around 78% for male. It also topped out at around 60% (so basically 15% increase, max) with the rate for males dropping.
Right, except I acknowledged my initial statement was wrong because I was basing it off of household income when you were actually referring to personal income. So why did you spend the effort refuting that?
Median personal income is per individual, which is obvious. After I corrected myself the question then became "did the typical individual's real income keep pace with the cost of the things they can't avoid buying?". The answer is no. And I already showed why.
I would assume being profitable constitutes as winning when you're throwing hundreds of billions of dollars at a technology. OpenAI and Anthropic don't have a very obvious path to profitability.
The author is pretty obvious and exhaustive about what he means by "losing": AI capex bubble is unsustainable, AI revenue is circular, no meaningful AI compute demand outside of OpenAI and Anthropic, AI products are mediocre at best (and still heavily subsidized, at that), AI is causing various mental health crises, OpenAI lost $20.9 billion on $13 billion in revenue in 2025.
OpenAI spent $17.2 billion on Azure in 2025 making the infrastructure bill exceed total revenue by about $4 billion BEFORE even counting salaries, research, stock compensation or anything else.
I feel like the responses here are purposely obtuse and people are refusing to realistically evaluate the economics of Anthropic and OpenAI
> The author is pretty obvious and exhaustive about what he means by "losing": AI capex bubble is unsustainable, AI revenue is circular, no meaningful AI compute demand outside of OpenAI and Anthropic, AI products are mediocre at best (and still heavily subsidized, at that), AI is causing various mental health crises, OpenAI lost $20.9 billion on $13 billion in revenue in 2025.
So then it’s just profitability modulo “various mental health crises”
Obviously it's not. Would you like me to sit here and do the work of reading the article for you? Or are the numbers involved a bit too much for you to handle?
Continue shoving your fingers in your ears and closing your eyes to the absolute nonsense that is behind OpenAI and Anthropic's economics, it won't change the fact that their path to profitability doesn't exist.
But Anthropic would be profitable if they stop training cries
No seriously, I recently heart someone say that and it may even be correct. However I assume this only to be correct if a stop in training doesn't change the current revenue environment - which it absolutely would. I am still convinced a lot of demand for tokens is artificial in the sense that people buy tokens because thats what you do right now (and its convenient) not because the roi is so great.
I guess I'm just confused by this sentiment. Are you making these conclusions while considering the fact that AI is still heavily subsidized? The economics of AI isn't quite the same as other software/tech.
I don't think it's going anywhere, but I don't know what happens when prices start to rise because these companies need to start turning a profit.
I'm happy using GLM 5.2 at API pricing. There will never be a point where essentially unlimited usage (i.e. what I, as a normal developer, can use at peak effort) of extremely good models will become financially unattainable.
> There will never be a point where essentially unlimited usage (i.e. what I, as a normal developer, can use at peak effort) of extremely good models will become financially unattainable.
The way you phrase this prevents refutation. But there will be a point where ordinary individuals cannot participate in the majority of the upside.
In the future, some class of models will require enormous compute that is outside the financial capital capability of ordinary individual contributors, middle class, and upper middle class. This will be sold as a capability to well-funded companies.
Yes, I don't disagree with what you're saying. I'm responding to a specific claim: that the capabilities we have now will become much more expensive in the future.
reply