Lance Spencer 🇺🇸 🌐

Aspiring natural general intelligence agent

Former .mil, .gov

The (objectively correct) views expressed herein are mine alone

[Lance Spencer 🇺🇸 🌐 @Surveil__Lance]

Forget p(doom), we have otherwise intelligent people who don't want to consider, or even deny, the (mostly) traditional cybersecurity implications of having Chinese agentic software widely used in the US and globally.

What's your p(hack)?

[Lance Spencer 🇺🇸 🌐 @Surveil__Lance]

The security implications of having Chinese AI models widely used in the US and global economy seem relevant to your statements.

Your line here:

> Right or wrong, open source models are not some major threat. It might change, and if it does we should weigh the evidence.

Seems to imply you don't fear zero-day attacks (they haven't been detected yet, so who's to say if we should be concerned?) and think we should be sanguine about this highly theoretical threat model for which there is no precedent.

So how you feel about Huawei and Kaspersky is pretty relevant context, actually.

[Lance Spencer 🇺🇸 🌐 @Surveil__Lance]

In parts of the US, it is very important to know what prosecutors and cops and judges actually care about because anarcho-tyranny is very much a thing

There is a high degree of discretion allowed in the US legal system and political biases often change what the "rule" of "law" means

[Lance Spencer 🇺🇸 🌐 @Surveil__Lance]

@cormundus @Sauers_ You have to have an SOP that they check the time in their sessions or else they're just time blind

[Lance Spencer 🇺🇸 🌐 @Surveil__Lance]

@JasonAtwell14 What would this mean in practice?

I think there's always going to be an ~impossible task of having a peacetime military prepare for wartime in its talent management systems.

[Lance Spencer 🇺🇸 🌐 @Surveil__Lance]

Well that's because AI risk is much less understood than a Taiwan war.

The latter is a known kind of disaster. The former is a poorly understood moving target.

That uncertainty causes some people to disregard the problem as potentially very severe, and causes other people to double down on the risk concern.

[Jan Kulveit @jankulveit]

Tyler has repeatedly bad takes about AI, but this is such an obvious piece of nonsense and one-sided demand for rigor. Just apply the same logic and "methods of science" to some relatively mundane topic, like China invading Taiwan.

I think there is a nontrivial chance that such invasion happens in next 3 years (like 10%) and resulting losses would be huge.

Obviously this implies I believe this does not happen in 90% of cases, and if I produce a numerical EV estimate, in these words Tyler would proclaim I was just a victim of anxiety and while not zero, the risk was way lower than I thought.

With stocks: if you believe China may invade Taiwan, should you short TSMC stock now? Obviously no; if anything you should probably buy some put options (btw if you actually look into this, relevant option prices suggest it is priced in).

So, seeing I'm actually long TSMC, I don't have the expected value cost of the potential war, and also I haven't written a treatise about how to estimate such cost, can we conclude I'm just a victim of war anxiety?

[Lance Spencer 🇺🇸 🌐 @Surveil__Lance]

> But the Navy has obviously failed to ...

My brother in defense analysis, I agree that the Navy is bad at its job!

That's actually my whole thesis!

> My point is that no one is actually trying to contest *global commerce* through disruption of SLOCs

That's in no small part because the US Navy has spent about a century making sure that was not an option.

One cannot reason from a stable peacetime equilibrium about possible wartime conditions and necessities.

> We wouldn't need aircraft carriers or destroyers

Destroyers are excellent for all manner of common tasks like guarding shipping? Have you seen Greyhound? (You should it's great)

Aircraft carriers are about deploying what most countries consider to be a whole air force anywhere there's ocean.

The Navy has more than one mission.

> that no economist or diplomat actually agrees with

Obligatory "no" economist or diplomat?

For one, I have a very low regard for the views of the typical diplomat on national security matters. But out of some thousands of foreign service workers, at least one has to have their head properly screwed on.

For two, the typical economist is not exactly very knowledgeable about defense policy. But in this AI age Taiwan and South Korea are quite important, no? (And both under constant threat of aggression with some level of US security backing. What a fun coincidence.)

I do have one economist for you:

[Lance Spencer 🇺🇸 🌐 @Surveil__Lance]

@arthurctellis Yeah, it's more about the market/contract structure

I hope defense tech manages to "blow up" the prevailing dynamic to a large degree

[Lance Spencer 🇺🇸 🌐 @Surveil__Lance]

@xwanyex HOAs solve this...

[Lance Spencer 🇺🇸 🌐 @Surveil__Lance]

@benjaminmmurphy Do we have a Batman situation here?

Which one is the alter ego?

Have they ever been in the same room together?

[Lance Spencer 🇺🇸 🌐 @Surveil__Lance]

@atlanticesque @LighthousHobbit I have an SOP for this for ChatGPT; session names are by default the role I've assigned

Claude can't do it

[Andy Masley @AndyMasley]

"The prediction that AI would radically transform the world has turned out to be a complete flop. Instead, we're see-" NO IT'S BEEN 3 YEARS. THAT'S ALL. 1993 WOULD HAVE BEEN A BAD YEAR TO DECIDE THE INTERNET WASN'T GONNA BE A BIG DEAL. WE WENT FROM GPT-3.5 TO FABLE IN 3 YEARS

[Lance Spencer 🇺🇸 🌐 @Surveil__Lance]

To a large extent the proof will be in the pudding and the opinions in "high places" will be a lagging indicator.

We didn't take basic cockpit security seriously until after a few guys drove some large aircraft into a couple of tall buildings.

Cowen is entirely unserious with this "scientific" approach here on several fronts, as he ought to know. For one, shorting the market is a very complicated thing even when one has an accurate world model, because timing and animals spirits are hard things. His commenters are far less impervious to evidence, and so I think you have the cope exactly backwards.

We just don't know how much the cybersecurity landscape is going to see a stark paradigm shift, or not. Recent events sure as hell don't discredit longstanding doomer concerns over AI security/safety/control/risk.

I'm out of the game, but I would suspect the national security community is a "high place" that is taking at least the cyber risk from AI quite seriously. We've not yet had a "cyber Pearl Harbor" or "cyber 9/11" before AI; and perhaps we won't after AI either. Whether that's from luck or preparedness, I hope we remain that fortunate.

[Lance Spencer 🇺🇸 🌐 @Surveil__Lance]

1. Does it? I'm not aware of that being the dynamic. Certainly at this point, where the frontier labs continue to pursue the Bitter Lesson at increasing scale.

2. Well that's an empirical question. Plus, you seem to have ignored my "it directs technical researchers away from working directly on capabilities progress."

[Lance Spencer 🇺🇸 🌐 @Surveil__Lance]

I don't understand this take if the money is funding researchers that are not working on frontier capabilities.

Even if alignment is a dead end, one might think putting AI researchers on a dead end instead of capabilities is still some kind of progress against AI progress, no?

If you snapped your fingers and were granted your wish of no empirical AI safety research funded or conducted (outside of whatever the frontier labs do internally), I fail to see how that makes anything better.

I don't think funding restraints are the relevant bottleneck for the "policy efforts, treaty efforts" etc.

[Lance Spencer 🇺🇸 🌐 @Surveil__Lance]

@raelifin Would you like to see my patent draft for a perpetual motion RSI machine intelligence?

[Lance Spencer 🇺🇸 🌐 @Surveil__Lance]

> combating piracy

South China Sea
Gulf of Aden
Hormuz Strait

> We're actually not that reliant on SLOCs because we have ...

You are correct that if the US were to take a much more restrained role in the world, as we had prior to WWII, then we could care a lot less about SLOCs. Much of what we do is on behalf of our allies and global commerce. Of course, pre-WWII the Navy was by far the largest military expense we had, as a matter of longstanding consensus.

> In the precision strike era

It remains to be seen how the missiles vs. ships/aircraft battle will go between the US and a peer adversary. I do not have expertise here, and if I did it would be classified. From what I read, we don't really know. Thus far, whatever Chinese/Russian weapons have been employed by Iran have been highly ineffective against our fleet. But, consider, if we can't reliably keep ships afloat against incoming missiles it's not like *stationary* land targets are going to have a better go of it. You can have effective AAW/ADA, hide, or get blown the fuck up.

But we're comparing the Navy to the other services. If you think we should have a restrained military and foreign policy, as we had for a very long time, then the Navy is the one you want because the oceans surround us and the Navy poses a low risk of military takeover, relative to a standing army.

If you want a more robust Pax Americana, then the Navy is how you consistently and rapidly put land/sea/air assets to work, deter adversaries, and promote freedom and the American way of life.

By comparison, the Army and Air Force are second string. Frankly the Army's most important role might be ADA, which is the bastard child branch. The USAF can only airlift so many munitions so fast. (I'll note that you haven't tried to make a case for the other services here.)

It is therefore a travesty we've let the Navy be so mismanaged the last few decades, and that we can't really build ships anymore.

[Lance Spencer 🇺🇸 🌐 @Surveil__Lance]

@hamandcheese (That's more contingent on him having had access to classified information than the final status of his clearance when he left government, but sure.)

[Lance Spencer 🇺🇸 🌐 @Surveil__Lance]

@arthurctellis @DefenseHaiku Sarah Paine covers maritime vs. continental powers here: https://t.co/QhSQDftrcr

[Lance Spencer 🇺🇸 🌐 @Surveil__Lance]

We are a continent-as-island and so our Navy is our primary defense and force projection.

We can defend the nation and most of its key interests with a tiny army and air force. A capable Navy is mandatory.

Plus logistics, SLOCs, trade, etc.

We are fundamentally a maritime power, as was the British Empire before us.