This weekly roundup thread is intended for all culture war posts. 'Culture war' is vaguely defined, but it basically means controversial issues that fall along set tribal lines. Arguments over culture war issues generate a lot of heat and little light, and few deeply entrenched people ever change their minds. This thread is for voicing opinions and analyzing the state of the discussion while trying to optimize for light over heat.
Optimistically, we think that engaging with people you disagree with is worth your time, and so is being nice! Pessimistically, there are many dynamics that can lead discussions on Culture War topics to become unproductive. There's a human tendency to divide along tribal lines, praising your ingroup and vilifying your outgroup - and if you think you find it easy to criticize your ingroup, then it may be that your outgroup is not who you think it is. Extremists with opposing positions can feed off each other, highlighting each other's worst points to justify their own angry rhetoric, which becomes in turn a new example of bad behavior for the other side to highlight.
We would like to avoid these negative dynamics. Accordingly, we ask that you do not use this thread for waging the Culture War. Examples of waging the Culture War:
-
Shaming.
-
Attempting to 'build consensus' or enforce ideological conformity.
-
Making sweeping generalizations to vilify a group you dislike.
-
Recruiting for a cause.
-
Posting links that could be summarized as 'Boo outgroup!' Basically, if your content is 'Can you believe what Those People did this week?' then you should either refrain from posting, or do some very patient work to contextualize and/or steel-man the relevant viewpoint.
In general, you should argue to understand, not to win. This thread is not territory to be claimed by one group or another; indeed, the aim is to have many different viewpoints represented here. Thus, we also ask that you follow some guidelines:
-
Speak plainly. Avoid sarcasm and mockery. When disagreeing with someone, state your objections explicitly.
-
Be as precise and charitable as you can. Don't paraphrase unflatteringly.
-
Don't imply that someone said something they did not say, even if you think it follows from what they said.
-
Write like everyone is reading and you want them to be included in the discussion.
On an ad hoc basis, the mods will try to compile a list of the best posts/comments from the previous week, posted in Quality Contribution threads and archived at /r/TheThread. You may nominate a comment for this list by clicking on 'report' at the bottom of the post and typing 'Actually a quality contribution' as the report reason.
Jump in the discussion.
No email address required.
Notes -
Unlike many people, I am definitely on the AI-train: this is a revolutionary technology that will transform the way we live. I do think that people and companies that learn to make use of this technology most effectively will gain compounding benefits over time, just like how it seems households that has a computer 30-40 years ago (either through wealthy parents and/or educated parents) really helped prepare their children for big and better things (although obviously having wealthy and/or educated parents already help a lot too).
If one looks at the AI Index of Ramp (corporate spend management platform), one will find that in monthly AI spend, the top 1% of companies are now hovering close to $10k/month/employee, with a big drop off that follows where top 10% of companies are only spending $800/month/employee, and the median company is only spending $15.70/month/employee. In short, being AI-pilled shows up on the balance sheets. In an interview with Derek Thompson (btw, the other author of Abundance) on his podcast Plain English, Ramp's lead economist Ara Kharazian says: "All of the companies in that top 1% are extremely technically advanced. They are concentrated in the information sector, so software and media or finance or technical services." I do believe these companies are on to something, these are evidentially smart collective of people with clear objectives (increase productivity, reduce cost, maximize profit, etc.)
On a personal level, after 1 year ish of using Claude/Codex as a software engineer, I've noticed many benefits of AI usage (though . I've shared before that it has freed up time, for example, to engage with this forum. I've also been on and know of multiple projects with which would have never even been greenlit if not for AI-use, such as integrating with an external partner on a legacy platform in a computer language no one in my company knows (tbf, it is a C-family language). There has been many complaints about software reliability in the AI-era, with one poster child being GitHub that got bad enough where prominent projects left for supposedly greener pastures. My company did the opposite in that we are migrating from Atlassian's bitbucket to github, I think in part because of what I summed up in an internal message at the time when asked of should we or not: "a migration to GitHub is essentially a bet on AI, because Microsoft/GitHub for sure will be at the frontier of AI features compared to Atlassian/Bitbucket." Regarding the perception of reduced software reliability in the AI-era, I fondly remember early Q2 this year, when the VP of Support and Operations came to my team and the CEO to complain about the increasing amount of outages; I sat down for about a day, counting up number of changes-features-fixes per release and number of incidents caused by each release before presenting to everyone in a meeting my findings: the rate of errors (number of incident / number of changes-features-fixes) actually reduced per release, but because we were releasing so often due to increased cadence and productivity, the raw number of errors increased. The aforementioned VP btw has come to learn to appreciate AI a lot, even if he doesn't quite understand the terminal still.
This brings me to the thread down low between @RandomRanger and @MaiqTheTrue which provoked this chain of thought, bolded mine:
I found a source for the stat btw.
Why is there more stress and suffering now that would lead to the tripling of teenage suicides? Ask the random person on the street, they'll just point everywhere and tell you to look around and say: "everything is so hard now". I think what is so hard about life in the US, and I guess the modern age, is complexity. Life these days are too complext. It's simple enough (heh) to look at computers as a technology that drives societal complexity. Computers makes certain problems extremely simple ("find the 25th, 50th, and 75th percentile using this column of data in Excel"), but then it leads to new problems that are more complex ("now that we know the 25th, 50th, and 75th percentile, what does that mean to us?"). Fifty years ago, it was simple enough to understand your own finances, at most you had a bank account, paper checks for the monthly payments on car and your house, and pay in cash for everything else. Now consumers look at managing multiple credit cards, point rewards and loyalty programs for various companies, managing their own IRA/401K/brokerage, etc. Fifty years ago, you marry the person you met in school or the person you met at work, now gotta navigate dating apps, hookup culture, shifting and varied interpersonal relationship expectations, etc. The following statement from @FCfromSSC points to it well:
And of course btw, complexity is a subsidy for the rich, because only the rich can afford to navigate all the complexities.
So what will AI do to/for us? Personally again, I think AI has greatly simplified my life. There are many questions that I wouldn't even know where to begin to ask, now I just ask. But a colleague has recently confided in me that he thinks of quitting before the year ends, "the rate at which we're going, it's just too fast". I was perplexed at that because I thought AI would have made his life easier like AI did mine. But maybe it wont? what do you think?
Sorta off topic, but since we're on the topic of AI, a good heuristic I have found is that predictions of bubble/crisis/unsustainably tend to almost always be wrong. The 2000 dotcom bubble or 2008, where high valuations led to sudden corrections, are outliers. Space-X is still above its IPO price of 135 despite fears of insiders selling and a bubble. I predict all three companies, Space-X, Open AI, and Anthropic will sustain $1-2 trillion valuations each without it collapsing. The media will be wrong as always, just as they were wrong about Amazon, Uber, Tesla, and many others. Those companies went from losing lots of money and being overvalued to being hugely profitable and much less overvalued.
AI coding and math are the killer apps. The reason why these are trillion dollar companies is by pivoting to coding. Every major company is using AI for coding.
Better tracking plays a role too. Record keeping and classifying causes of death has gotten better now compared to 50-70 years ago. Also, more ways to commit suicide, like pills, which may not have existed in the 50s. I 100% agree things have gotten way harder (unless you're smart and talented enough to land one of those high paying tech jobs) , but I think it's a leap to say this has led to more suicides. Complexity is part of the problem, but also more competition. America produces a lot of smart, ambitious people vying for relatively few slots--be it good paying jobs or credentials. People must miss out.
The AI bubble predictors have already proven correct, just...in China.
Setting aside China, I'm not sure that your heuristic here is correct or that looking back over the past couple of decades is the right period to be searching. Besides the dot-com bubble, past situations analogous to AI would have been the railroad and perhaps electrification, and in both cases there were bubbles that popped (the second intertwined with the Great Depression).
One of the ways in which I think that AI is not like past bubbles is the fact that the technology is not yet mature. I don't mean this in an "AI is bad way" but rather that it's sort of unknown what the "meta" will be. New advancements might make running models on your home device easier and hurt the primes, or it might result in newer and powerful models that only large organizations can run. With railroads, the internet, and electrification, even though there was a great deal of technological advancement left before they were laid down, the general shape the technology would take was fairly well established in a way I am not sure AI is. This could be good, or bad, but I do think it makes prediction sort of fraught.
China does have structural disadvantages in AI though. They need to build out an entire hardware and semiconductor stack while competing with the US that has a huge lead in both quality/quantity of AI hardware. Spend 500 billion to be in a mildly less distant second place... not very profitable, even if strategically necessary.
That's reflected in the index in that article. The Chinese STAR50 index is made up of early-stage 'competing to unseat the established US market leader' companies that are innately risky but potentially highly lucrative if they can actually do what they claim. They're supposed to have really high valuations compared to earnings because they've only just started making profits. They're supposed to be risky too.
The US equivalent would be an index tracking recent AI startups, neoclouds and semiconductor midcaps, beleaguered national champions like Intel that are supposed to be turning it around and runner-ups like AMD. I think the article makes an apples-to-oranges comparison between the NASDAQ and STAR50. The NASDAQ tracks the incumbent superheavyweights like Nvidia and Google, its p/e is far lower than China's 'post bubble pop' STAR50. They're not the same kind of index at all.
Consider US or European carmakers facing the Chinese onslaught. How do they feel? Outnumbered and outgunned. Does it really make sense to invest a hundred billion in modernizing the car industry when you're facing a foe with gargantuan industrial capacity, who's already modernized and is still far bigger than you are? Maybe, depending on tariffs. But not exactly promising for investors, they won't be that excited about it.
I agree with a lot of what you are saying here, but I think there's a few things worth addressing. The first is that there's going to be massive domestic demand within China for a domestic AI, because China is cognizant (as you point out) of the necessity.
The second is that open-weight models have been having a moment recently, which indicates that there's a future for Chinese models even in a world where the best models are American.
Finally, I'm not sure you're right about the STAR50. AMEC and SMIC don't match that description (they are decades-old chip fabs, SMIC being the largest in China), and those are just a couple that caught my eye.
SMIC and AMEC are older but they're kind of like Intel, beleaguered national champions facing serious competition. They're not masters in their field, they have to do all this investment and they can't reap profits because they're not on top yet.
Intel forward p/e is about 60x, same as SMIC and AMEC. Meanwhile TSMC is about 23 and Nvidia somewhere in the 20s.
Obviously, nothing is exactly alike between the two countries. They have different market dynamics, companies like Bytedance have near Facebook revenues and profits but much lower market capitalization...
China sure does need compute but they can't really get it, so they have to make highly efficient but less capable models. Their opensource models do well on some marketplaces like Openrouter or Vercel but those are a small portion of global AI spending and tokens. Most demand goes to the cloud providers directly, through webapps or Codex, Cursor, Claude Code... There's no need to go through these middlemen companies. Opensource is a significant part of the market but it is hard to profit from, since anyone can run it internally.
It's hard to make a business case for an AI company being worth hundreds of billions if it can't capture the profits from its own products, if Nebius or Fireworks or whoever is making that money instead.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link