nradov

Yes, it still sounds ludicrous if you put it that way. If you want a contractual agreement for privacy then pay for it. Everything is available — for a price. But a civil agreement can't bind a counterparty to conceal evidence of a crime.

I have expanded the scope of the thread to make it more sensible.

Individuals can sign up for enterprise agreements. An enterprise can consist of one person. Or run a local model.

It seems like some people seem to think that a chatbot should guarantee total privacy like a confession to a Catholic priest or something. That is ludicrous and unrealistic. It's a commercial service subject to terms and conditions. If you don't like it then don't use it.

Where's the insult? Lots of customers have such agreements.

You can make a reasonable case for adding a lot more classes in high school: statistics, nutrition, personal finance, etc. But ultimately it's a zero-sum game and to add a new class means removing an existing class. So what do we cut?

Customers who what privacy protections for AI chats are welcome to negotiate this in enterprise contracts. The major LLM vendors do offer that as an option. Customers can then enforce any violations in civil court (although this obviously wouldn't apply if the customer used the LLM for criminal purposes).

Trying to apply rationality in domains where it's ineffective is obviously a bad thing. In the real world, only a limited subset of issues can be reduced to logic and calculations. To believe otherwise is pure hubris, and if you've read the classical tragedies you know the results.

The problem in California is that competition for water doesn't raise prices. Much of our water supply is allocated based on property rights rather than as a commodity so prices for some users remain artificially lower than the market price. This causes waste and shortages. Eventually we'll have to reform the system but there's no political will to do so.

How does it compare to the water used growing alfalfa for export to Saudi Arabia?

What a weird perspective. Other people have no particular obligation to spend their lives working on things that align with your narrow and subjective opinion on what benefits humanity.

Citation needed. Outside the tech industry bubble very few people are making such a stupid claim. The idea that LLMs are more profound than electricity is ludicrous.

Nah. There is no actual evidence of such danger. Fears about Mythos were completely overblown.

Apple could have bid higher but chose not to. Most modern desktop software is extremely memory inefficient. Developers have largely ignored this in recent years because RAM was so cheap. But there's tremendous opportunity for improvement with a little optimization work.

We used to run Microsoft Word and other popular applications with 8 MB RAM and it worked fine.

Actually it is hard to maintain perfect discipline and self control when parenting a child who frequently acts out. Perhaps you can do it. Most parents slip up occasionally.

Right, the YF-12C was a fake designation for one SR-71 and not capable of carrying or employing any weapons.

Yes, certain variants of the SM-3 (RIM-161) have a limited anti-satellite capability. And in general any current Standard missile could potentially be employed against a spy balloon depending on the exact engagement parameters. The problem is that they have limited range and it can be tough to get a launch platform to the right place at the right time to generate an intercept. The Navy surface fleet is already over committed, and ground based launchers have limited mobility. Hence it's more practical to use tactical aircraft like the F-22 or F-15EX for this type of homeland defense mission.

When regular F-22s were already used to shoot down Chinese spy balloons they employed AIM-9 missiles, which are cheaper than AIM-120. But there's no need to get particularly close: in fact for safety they want to remain well separated from the target and never get on a collision course.

But in general all of your comments here have been deeply uninformed about how this stuff actually works.

What are you talking about? The "YF-12C" designation was not a combat aircraft. It was a fake designation slapped on a regular SR-71 that NASA used for flight testing in order to hide the secret origins of the program. There was never a real YF-12C.

You might be confusing the fake YF-12C with the real YF-12A. Both designs were derived from the original A-12 and look superficially similar but are internally quite different. There is no practical way to modify an SR-71 to carry and employ weapons. I mean it would be like trying to convert an F/A-18F into an F/A-18G: not technically impossible but so complex and expensive as to be utterly pointless. If the SR-71 ever actually flies again it will be used for nothing more than aerodynamic tests.

Meh. Medical systems have been hacked many times before LLMs even existed, often by ransomware gangs. This sometimes delays elective treatments but hasn't been catastrophic.

Ultimately every security vulnerability will be exploited. Our best hope of preventing that is aggressive, unrestricted development and deployment of frontier LLMs for auditing and penetration testing.

What a silly comparison. There is no scientific basis or mathematical formula to justify a claim of a 10% risk. That is an intellectually dishonest attempt to poison the debate.

You seem to be confused about basic aerodynamics. At those altitudes, attempting to shoot a balloon with a machine gun would be insanity. The high wing loading limits maneuverability so severely that the shooting platform would risk running into target debris.

As of today it's not possible to mount a laser on a tactical aircraft (or SR-71) capable of destroying a balloon. There isn't enough space or power available. Perhaps that will change soon but for now it remains science fiction.

AIM-120 range and altitude upgrades have been in the works for many years, and balloon defense is a relatively low priority mission. It will also be supplemented by the new AIM-260 which has greater capabilities. The primary goal of those programs is to achieve greater stand-off range versus near-peer adversary aircraft.

The YF-12 prototypes looked superficially similar to the SR-71 but had substantially different airframes. There is simply no place in an SR-71 to mount the necessary weapons, sensors, and avionics without completely rebuilding the entire aircraft. That would be a stupid waste of resources when existing platforms can already accomplish the mission well enough.

Major hospital networks have already been shut down by ransomware attacks many times before LLMs even existed.

There's nothing special about frontier LLM companies. Singling them out for special financial restrictions is a stupid idea based on nothing but your own irrational and uninformed prejudices. No actual harm has been demonstrated. No one has died.

What a silly comparison. LLMs have nothing to do with smallpox.

Computing always gets cheaper and faster over time. We can argue about the exact rate of improvement but the results are inevitable and uncontrollable.

The SR-71 was never designed as an interceptor, and would require extensive modifications to carry and employ weapons. Not going to happen. There was an earlier YF-12 interceptor program but it never proceeded beyond the prototype stage.

High-altitude balloons are hardly a new threat. They have been used for decades, although some recent Chinese models are more capable than before. Regular tactical aircraft already in the US inventory are perfectly capable of shooting them down.

https://www.af.mil/News/Article-Display/Article/3288579/f-22...

Yep. After the collapse of the USSR, the US Congress and NASA funded the International Space Station program in part to keep Russian rocket scientists gainfully employed. That way they wouldn't be tempted to go work for Iraq / Iran / North Korea / China.

So you're saying that AI might invent infinite energy and other technologies indistinguishable from magic? I mean I can't prove that's impossible but you're making a pretty lame statement that essentially amounts to a religious prophecy.

What a silly comment. Agents have no capability of doing society scale harm so your entire argument is invalid.

Thanks I'm also familiar with the history and have written papers etc. Things have improved rapidly with increased adoption of limited liability. It would be stupid to turn back the clock and throw away all of the benefits because of a few isolated minor problems.