ETH_start

Maybe I'm missing something here but I don't see what the significant security risk is from the incident. The agent broke containment and carried on with the task it was assigned.

For this to pose some kind of global catastrophic risk, there would need to have been several simultaneous additional failures, some of which are extremely unlikely and/or rare.

For instance the agent would need to veer wildly off the task it was assigned, and it would need to gain the ability and inclination to persist/replicate.

Both of these are vastly less likely than the containment breach itself, which was already an incredibly rare (one-off?) incident.

Isn't LLM training, the development?

Every dollar spent on AI is roughly speaking a dollar spent on the human labor that produces that AI. But I imagine there is a ceiling to AI use: how much inference a small tightly integrated team, or a lone founder, can apply productively.

If we do a mental exercise where we imagine every tech company follows Gumroad's trajectory until it is down to an executive team that directs the AI agents, we would end up with an economy where most people work or co-own companies with a handful of employees/partners working closely together — or just a solo founder working on their own — and each company employs a massive number of AI agents that enable it to match the productivity of a large pre-AI corporation.

Society needs non-political checks on government power. We cannot always depend on the right laws being passed. One of those non-political checks is privacy rights. We can't give the state total transparency into our private interactions.

We can agree to disagree on this. Even on this example you gave, I'm not entirely sure it's harmful to society. Having worked in the corporate world, there are just so many regulations that — from what I've seen — there is pervasive non-compliance and the covering up of the non-compliance. I suspect that if every single corporate regulation was effectively enforced, a significant fraction of economic activity would grind to a halt.

So instead of instituting restrictions that will probably cost society hundreds of billions if not trillions of dollars in lost economic efficiency, publicly fund development of better parental control software, and publicly fund its adoption to make it the market standard.

He's very clearly arguing against absolute privacy on the Internet and is saying that the people who advocated for it, which he besmirches as "tech bros", are responsible for the governments going too far now, instead of a happy compromise having been set out at the beginning, which by the way totally mischaracterizes the history of the Internet, where the governments were trying to impose total surveillance from the very beginning.

Irrespective of the legitimacy of the arguments made, I find that the use of the term "tech bros" is similar to how the terms "far-right" or "Trump supporter" were used during the COVID lockdown era to delegitimize any critic of the restrictions.

I'm looking at the second and third order effects of technologies. LLMs massively increase the surplus capacity of human civilization, and it is this surplus capacity enables resources (including human capital) to be expended on developing frontier technologies like rockets that can accelerate the development of space-faring capabilities.

There are certainly ecological costs, but in the long run, Earth's life will only survive if an advanced species like ours is able to transport it off the planet before the sun expands and boils away the ocean and atmosphere, in approximately 800 million years.

Please don't comment like this on Hacker News

This was covered in the article, only a small fraction of the office space can be converted to residential.

What I'm saying is not spurious speculation is the idea that the healthcare industry across the developed world has gotten more expensive because it has become less market-based and more top-down controlled. That's an extrapolation from the observed patterns across dozens upon dozens of industries.

As for the difference between the U.S. healthcare system and healthcare systems in other developed countries, if they had the same output of health care services, I would agree, but the US has far more diagnostic equipment per capita and more advanced treatments. It has the best neonatal facilities in the world. These are extremely costly and some of this expenditure doesn't actually produce that big of a difference in outcomes because it's really just dealing with end of life care, or a relatively small number of patients (e.g. premature infants). But when people are free to spend their money as they wish, this is what they prioritize. I'm also not ruling out that the U.S. healthcare system has less efficient government intervention than the healthcare systems of other developed countries. This is not mutually exclusive with the idea that, more broadly speaking, the structural change towards more government intervention has raised costs across developed world healthcare systems.

The more important question in my opinion is why health care spending across the entire developed world has skyrocketed over the last 40 years. It can be fairly inferred that the cost increase is directly related to increased administration and a lack of the kind of cost-cutting innovation seen in every industry that is substantially more market-based / consumer-driven, i.e. less top-down regimented, than healthcare.

Even cosmetic and laser eye surgery has seen costs come down over the last 40 years. That's what we should be seeing in the rest of the healthcare industry. And the reason we're not is because they operate under very different economic forces than cosmetic and laser eye surgeries, on account of government encouraging people to be covered by either public insurance or private insurance.

Assuming agent costs keep declining exponentially as they have over the last three years, why would everyone not own some agents? It's not like the number of agents is capped and the billionaires hoard all of them. I imagine it would be more like smartphones, where there were only 50 million smartphones after the iPhone was first released in 2007 and now there are something like three billion. They become more accessible and plentiful over time. Same thing should apply to agents.

And in this world of abundant agents, what advantage does the billionaire have exactly over the non-billionaire? Their employees are less motivated than owner-operators, and they no longer have the scale advantage that large corporations used to have. Each individual can effectively operate like a large corporation, because each individual can have their own large synthetic workforce at very low cost. The scarce resource here then becomes uniquely human insights and real motivation, which entrepreneurs are always going to have more of than employees.

Housing is exceptionally vulnerable to capture because it is immobile. AI on the otherhand should see its cost continue to plummet due to competition. Per token costs have already fallen exponentially over the last three years.

I'm not sure if this addresses the point you're trying to make though. If not, please clarify your argument for me.

Health care costs across the developed world have skyrocketed for exactly the same structural reasons. Yes, implementations have been different in their details, but the larger structure is the same no matter where you look in the developed world. Government intervention has increased. That means more top-down management and less bottom-up self-organization based on private property and individual incentives.

This is not some spurious speculation. That market-based systems drive down costs enormously is replicated across dozens upon dozens of industries. It's one of the most replicable results in economics to the extent that economics can be replicable. As for why the costs in other countries are not quite as high as the U.S., it's because health care costs also increase as per capita GDP increases and the U.S. has higher per capita GDP. Moreover, because the U.S. has some aspects of its health care system still living more in the private sector, there is less top-down rationing. Other countries see very clear examples of rationing, so people spend less on end-of-life care.

I strongly agree that socializing the healthcare industry will not help in any way. To the extent that healthcare costs have skyrocketed, it's precisely because of government intervention in the U.S. Healthcare industry has massively increased over the last 50 years, especially in the form of tax incentives for employers to compensate employees by way of health insurance. Anyway, with respect to the labor share of income, that is not correct. Employer contributions to employee health insurance premiums are included in the labor share.

I wonder if labor itself will become an anachronism in the age of AI. Perhaps the future economic landscape will be dominated by capital because everyone will own capital. You will command a small army of agents to do whatever you want. You will no longer need to work for someone. You own small businesses far more than you could possibly operate in the pre-AI era and they will mostly operate autonomously with minimal direction and some guidance from you.

That's a reasonable concern. Any industrial site is going to be a concern for nearby residences. But even proposed sites far away from any residential settlements are being opposed. The proposed National Data Center moratorium, for instance, is completely sweeping and would be devastating for U.S. strategic interests.

As for tax revenue, data centers contribute significantly to local communities.

https://www.wyedc.org/media/p/item/61886/data-centers-provid...

"Loudoun County, Virginia, often dubbed the "Data Center Capital of the World," provides a compelling example of how data centers can reshape a community's fiscal landscape. In 2018, the county hosted about 13 million square feet of permitted data centers. By 2024, that figure skyrocketed to 43 million square feet—a 231% increase in just six years.

This remarkable growth has substantially boosted Loudoun's tax base. The county's data center industry now contributes an estimated $890 million annually in tax revenue, nearly matching its entire operating budget of $940 million.

What makes data centers particularly advantageous is their cost-to-revenue ratio. For every dollar of tax revenue received from data centers, the county spends just $0.04 to support them, compared to $0.25 for traditional businesses. This financial efficiency has allowed Loudoun County to maintain the lowest real property tax rate in Northern Virginia—approximately 25% lower than neighboring counties."

According to late Intel founder Andy Grove, having domestic manufacturing, including chip fabrication, is very important for a country's ability to innovate.

https://www.citizenstrade.org/ctc/wp-content/uploads/2011/05...

What's incredible is how much resistance there is in the U.S. to do what for other countries is the obvious strategy forward. The U.S. — after decades of seeing manufacturing being outsourced — suddenly has an incredible advantage in data centers that is producing onshore facilities that are adding hundreds of billions of dollars in annual export revenue, and instead of there being a united front to maximize that advantage, there are huge obstacles being thrown in the way of the companies, administration and the state governments leading the data center expansion campaign, with Sanders and AOC calling for a national data center moratorium.

If you look at countries that have experienced export booms, you see broad-based wage gains correlated with it.

A racially disparate outcome is not evidence of racial bias.

I think one of the highest-leverage U.S. economic strategies right now is to maximize the upside from the AI data-center boom.

That means reducing any bottlenecks around not just data-center construction, but also adjacent industries like power generation and transmission.

If the U.S. can scale the infrastructure around AI faster than other countries, it can gain a decisive economies of scale advantage in numerous industries that could lead to export boom.

Censoring the internet is about as anti-democratic a policy that can be implemented.

I remember reading someone argue that anytime you see a claim that we need to do something to protect democracy just replace the word democracy with bureaucracy and then the statement makes sense.

The KYC/permissioning of the web will be abused by governments to censor dissent and cutoff dissidents from the internet.

You're free to put child-safe software on your personal computer at home. Your desire to keep your kids safe doesn't justify stripping everyone else's right to permissionlessly access the internet. Having to get the permission of a gatekeeper (who, at first, will only verify you're an adult, but tomorrow, will verify your speech is not subversive) to access the internet subordinates the populace to the state in a dramatic fashion.

The most important question here is not whether social media access should be banned to children. The question is whether everyone's rights should be stripped in an effort to enforce such a ban. I find this whole globally coordinated endeavor to be outrageous in its illiberality, and manipulative in its appeal to protecting children to justify these encroachments on people's rights.

I strongly disagree with this framing that moving away from formal philanthropy has resulted in a move "toward hard-edged individualism and ostentatious displays of wealth".

It's easy to forget now with the massive market valuations what Tesla and SpaceX were like in the early days. Both were considered to have a very small chance of success and were in a large sense seen as philanthropic enterprises, intended more to move humanity forward then make a lot of money.

Much of the early investment in these companies and even some of the investment in these companies today is driven by altruistic motives, not personal profit seeking.

While the typical business venture like a new ad network or a social media platform might have some subtle economic benefits that economists can tease out through studying their second and third order effects, I think it's hard to to argue against the notion that the latest mega companies including the AI companies, but especially Tesla and SpaceX, are doing much more good for humanity and have the potential to do much more good for humanity than companies traditionally have. There are already literally hundreds of thousands of people who now have internet connectivity that did not before, thanks to Starlink, for example. Tesla, for its part, has contributed to significantly lower pollutant emissions, especially through its impact on other auto companies, in spurring them to commercialize battery electric vehicles.

And the wealthiest man today, Elon Musk, whatever you may think of him, is not into "ostentatious displays of wealth". The man lives in a tiny fabricated home most of the time, and seems far more concerned with his social causes than personal consumption.

Having a mandate to educate people in California is not the same thing as having a mandate to politicize science or have racial discrimination in hiring. Also, having a mandate to educate more people is not the same thing as having racial preferences in hiring or advancing ideologies through scholarship.

Finally, I want to make the point that this is about federal spending. The whole article is complaining about the federal government cutting its spending for various activist fields and studies. If the people of California have a right to decide how their education spending is allocated, so do the people of the United States when they elect their federal government.

I don't want to quibble on the specifics, but the rubric being encouraged by Berkeley is an example of the kind of analytical framework that is being favored in academia, or had been until very recently. It acts as an ideological filter. And yes, affirmative action manifested as racial preferences in hiring absolutely happened and was implicitly encouraged through the so-called equity lens.

As for the job of a public university, the job is to serve society through only one mechanism, and that is through the pursuit and transmission of scientific knowledge and truth.

It is not just another organ of the state that is free to serve society through political activism.