Look if we are going to raid pensions and 401Ks to prop up the valuation targets of Anthropic and OpenAI longer, then the bubble can stretch further and further, but eventually datacenters have to get built and powered. It's the power generation part that no one talks about. We simply don't have enough power in the United States to scale at the rate that Anthropic and OpenAI need to prop up their absurd valuations. AI is real, but these valuations are not.
TL;DR: Ed is directionally correct, but it's anyone's guess as to the exact timing.
In the meantime I'm not going to complain about subsidized credits from the big labs. :-)
So US manufacturing would go up and up by that logic. Not to mention the US can simultaneously restrict the money supply via increased taxes at same time. The US also could just stop issuing bonds entirely and simply create the currency directly rather than the Rube Goldberg that is the bond market
I'll tell you what is going to happen, b/c it is happening as we speak. The U.S. Government is soft-defaulting on the debt by devaluing the currency. (The debasement rate is somewhere in the 7%-8% range)
There will never be a real default, but it is likely that the USD will lose reserve currency status. The U.S. Government has $114+ Trillion in total debt. (Something like $325,000 per person in the US) We are never paying that off. The only way we can do anything about it is to grow the economy and devalue the debt via inflation.
To all the other currencies that are currently used as reserve currencies, other than the US dollar. There has never been a single reserve currency in the world.
It doesn't matter who/what gets it next, what's important is that we're about to take a 60-80% haircut on our standard of living. It's what happened to the British when the lost it with the Pound Sterling, and now it's our turn.
Most people don't remember what hard money is, and why it's important.[1]
* The value of the U.S. dollar against other currencies dropped about 11% in the first half of this year, the biggest decline in more than 50 years, ending a 15-year bull cycle.
* Morgan Stanley Research estimates the U.S. currency could lose another 10% by the end of 2026.
* Despite a recovery of 3.2% in July, the delayed impact of tariffs on growth and unemployment – besides policy uncertainties – are likely to keep negative pressure on the dollar.
* Foreign investors have been adding hedges to their exposure to U.S. assets, which will likely further weaken the dollar.
World reserve currency was solved by Keynes and Joan Robinson in the 1940s.
What you do is have a currency no country controls that acts purely for exchanging one national currency for another. Give it to the united nations or something.
If you take a national currency and make it a world reserve currency then the nation that does do has problems with balance of trade, deficits, and mathematically screws themselves over. This is because you currency becomes overvalued compared to those of over countries, meaning your domestic manufacturing has problems competing, among other things.
We already made the mistake of one national currency becoming the world currency post WW2. I hope its not a mistake the world chooses to repeat.
This is what cryptocurrencies do, but remove the need for an intermediary and codify the rules in software, so it is much harder (near impossible?) for politicians to manipulate the money supply.
It will likely be a cryptocurrency or perhaps a basket of cryptocurrencies. You essentially want a way to store wealth directly on the Internet as it is the fastest and easiest way to reassign ownership of money or collateral. The nice benefit too, then is the U.S. (and no other country as well) is subject to the Triffin Dilemma.
Which was never really a problem before cryptocurrencies. Nor does it any way address the core issue we have with physical currency, taking out loans now, against future wealth generation, which is happening in the crypto space too.
Incorrect. It's always been a problem. Go back and look what has happened to the purchasing power of the dollar throughout the 1900s. If the dollar is losing purchasing power you can hedge that loss by buying Gold, commodities or cryptocurrencies. (Basically anything scarce) Also, there are no loans being taken out against the future wealth generation in crypto. That is entirely a fiat phenomenon. Call your Senator.
The EU Council are some of the worst offenders when it comes to annoying cookie banners. All regulation and compliance do is make things more burdensome for new entrants, which inevitably allows big companies to build moats that don't have competition. The side effect of passing such legislation was easily foreseeable and yet the EU did it anyway. There are legitimate reasons for companies to track and if you felt that the tracking was malicious, then don't do business with the company. The regulation was never needed.
What's very obviously unnecessary is the need for a law to police this. You can just not use things you don't like. This need to project one's own morality upon others will be the source of endless conflict.
What are your thoughts on introducing a law that requires social media sites with infinite scroll to feature an addiction warning, something like: "Using this platform may cause addiction in some users."?
There's good reason for a lot of the inaction around asbestos. If it's not moving, not being touched and not accessible to kids / pets etc then it's fine to leave alone. The alternative being extremely costly remediation, and there are still millions of structures with asbestos in them.
... Prop 65 is dumb though. Kind of a trend in California law.
This isn't useful advice for addictive things that have high short term reward and high long term regret though. Especially so when the other party has a strong incentive to keep you trapped in that loop of regret.
Exactly. Also, for responders, let's skip past the "if addicts didn't want to use them they wouldn't," and go straight to framing it as a moral failing implying that they deserve it.
I don't consider addiction a moral failing. But I do think it's wrong to place all the blame on social media if you also play an active role in sustaining your addiction.
That statement leans very hard on the word “like”. Do addicts “like” their addictive behaviors? In a way it doesn’t matter, they’re going to keep doing it anyway.
Conversely it isn't very cool seeing the person who grew up as a McLaren fan dying prematurely due to being influenced to start a deadly habit from a young age. And I say this as someone who enjoys smoking.
What is the cost-benefit analysis that led you to that conclusion? I ask as someone who enjoys smoking but struggles to see any good argument for the return of tobacco advertising.
The law being discussed is for minors. As a society we don't think minors have sufficient executive function to make those decisions, so we find it acceptable to ban them in those context.
Alternatively: As a society we don't think ALL minors have sufficient executive function to make those decisions, so we let parents who do make that decision.
This societal view that you are a powerless NPC slave to ones desires/addictions is highly toxic. It is not in fact binary for most people and people do have free will to make different choices. To absolve the individual of accountability is to undermine the very fabric of society. The way individuals typically overcome addiction is through reliance on close family & friends. Those family & friends can "take the phone away" or "uninstall the app" if there is really an issue. Laws are a blunt instrument and should be used sparingly.
I’m a Buddhist, so I really do see what you’re saying about personal responsibility and agency, but the other problem I see is that the companies at the forefront of these ultra-addictive industries (including mass social media) get busted eventually every single time for working overtime to hide what they knew about how damaging their products can be, and in fact to convince the public that their products do the opposite of what they actually do.
Does nicotine have some beneficial health effects? Yeah, it actually does. But do they outweigh the downsides (and societal costs)? Absolutely not.
I do believe strongly in individual freedom of choice, but only under a paradigm of informed consent, and companies like Meta have never allowed that to happen and they never will, because the truth getting out could destroy their bottom line.
> Does nicotine have some beneficial health effects? Yeah, it actually does. But do they outweigh the downsides (and societal costs)? Absolutely not.
What are the downsides and societal costs of social media addiction? We have decades of strong evidence on the harms of nicotine and cigarettes, but the evidence regarding social media addiction is nowhere near as clear-cut, at least from what I've read. Likewise, the causal relationship between cigarettes and their harms is well established, whereas the relationship between social media use and addiction-related harms is much harder to disentangle from other contributing factors.
Not to mention, nicotine is highly addictive for almost any user, while most people who use social media do not become addicted.
I understand the broader point you're making. I'm just not convinced there's enough evidence to justify treating social media addiction as comparable to cigarette addiction.
That article gives a superficial overview of some of the research, which for obvious reasons is still in its infancy.
I didn’t mean to compare the harms to tobacco, but to how these companies tend to operate.
That said, the fallout of social media addiction and overuse is far weirder than the fallout from tobacco (which mostly just gives people cancer). Social media appears to be tearing the fabric of society apart.
I don’t have studies to cite but it seems like half of americans are one type of conspiracy theorist or another now, and that’s directly attributable to these new media channels.
I think the negative effects have been most pronounced on young women (body image issues and bulemia) and people with low media literacy (the misinformation plague).
There’s also the algorithmic “extremism” pipeline, where anything you show a little interest in gets turned into more and more extreme iterations to keep you engaged.
So maybe “addiction” isn’t even the right word to use here. But just look around outside in any major city or event in America and see how compulsively almost everyone uses their phones during downtime. It’s reflexive at this point, not really a conscious decision.
Some of the scariest stuff is coming from primary school teachers. Most of them act like late gen z and gen alpha are genuinely illiterate. It makes sense, you have little kids trying to swipe on physical photos in picture frames and stuff. That says something big about their lifestyles and then imagine the downstream effects of that…
I think it really tells you something when the CEOs and people who work for social media companies don’t let their own kids use the product, or drastically limit access to it.
This is sparingly. The government basically let the tech industry run amok for 20 years and only acted when the problems it was causing were serious and undeniable.
Probably the most general-purpose one is SageMath, which is open-source and basically Python with a ton of sophisticated math stuff built into it. Everything I used to do in Mathematica I now do in sage, and I don't think I'm the only one. Probably field-dependent though.
Of course there is a whole constellation of more specialized things in certain fields, that has come a long way in the last 15 years. So people needing things like that no longer kludge things together in Mathematica.
Does SageMath use Sympy, or is there some other integrator built in? Last I heard Sympy was one of the worst performers, even among other open source CASs.
I thought distillation meant small models don't have to compete with the big models and can always eventually achieve close parity, but it's just a matter of time to do the distillation? (i.e. how much lag do you want to live with) Am I oversimplifying?
There is likely a theoretical limit to how much intelligence you can pack into a model of a given size (especially when stretching that over a large input context size).
Our evals are pretty complex so we only recently started testing ~30B class models, which are now becoming quite smart (on par with the frontier from 1 year ago). Mistral is far behind, but I'm rooting for them.
The reason why you will never get software engineers (in companies) to accept the man behind the curtain is liability. If a human software engineer is still responsible for what happens when the AI developed code has a catastrophic bug or security vulnerability, then the only way for the human to know if there is a problem is to be able to read through the code or run it through some <insert advanced formal verification tool here> that guarantees zero issues.
I think we eventually end up at the tool approach via vendors providing the tools to other companies, but it still feels like there's a long road ahead to get there.
TL;DR: Ed is directionally correct, but it's anyone's guess as to the exact timing.
In the meantime I'm not going to complain about subsidized credits from the big labs. :-)
reply