This is a special post for quick takes by Prudhviraj Naidu. Only they can create top-level comments. Comments here also appear on the Quick Takes page and All Posts page.
If military AGI is akin to nuclear bombs, then would it be justified to attack the country trying to militarize AGI? What would the first act of war in future wars be?
If a country A is building a nuke, then the argument for country B to pre-emptively attack it is that the first act of war involving nukes would effectively end country B. In this case, the act of war is still a physical explosion.
In case of AI, what would be the first act of war akin to physical explosion? Would country B be able to even detect if AI is being used against it? If there is an intelligent explosion, wouldn't the nature of war possibly change rapidly that country B can not even detect they are being attacked until it's too late. For example, you develop AGI and use that to hinder political process and cause slow down in economic growth rate by 2 percentage points. Over a long time horizon like a century, this dooms country B.
Then isn't it country B's interest to consider building military AGI as an act of war since it can not hope to detect future acts of war carried out by the AI.
Past History: Britain's first act of war against Germany in WW1 was cutting off the telegraph lines carried out by a non-military vessel. The first act of war in future wars does not have to be physical explosion and likely be an information attack.
Former Indian Army Chief speaks on how India is losing the Information War:
The "information war" sounds like politics as usual. Propaganda, censorship, Twitter, TikTok -- all have existed long before AGI.
International politics is not about fairness, but about how strong you are, who are your allies, and how far you can go before they stop supporting you. Israel can do whatever it wants, because there are many people in USA who will defend it no matter what. India does not have that kind of support. On the other hand, there is probably no need to worry about Trump; he always says some strong words to show his fans who is the boss, but the most likely thing he will do is nothing, unless there is some personal profit in it for him. (Hint: If you think you need his approval, bribe him. Not USA as a country, but Trump as a person; it will be cheaper and way more efficient.)
Seems to me that the distinction between "acts of war" and "technically not acts of war" is mostly in their deniability, which of course is a matter of interpretation. Bombs are difficult to deny. State-sponsored cybernetic attacks on other states' infrastructure? We pretend for decades that we don't see them, because... well, they are trivial to deny, you just need to play dumb.
Firstly, I want to say thank you for commenting on my shortpost.
The "information war" sounds like politics as usual. Propaganda, censorship, Twitter, TikTok -- all have existed long before AGI.
It's more about the magnitude and its effectiveness. For ex, India-Pakistan fire at each other on a fairly regular basis but that doesn't count as a reason to go to war. But if one of them drops a nuke, then the security of the other is endagenered to the point they will invade.
An AI / AGI powered "propaganda" machine can spin up millions of bots, argue with every individual in a country. Create personalized realistic images / fake news articles and news organizations that overpower information agencies (like actual newspapers). I don't think pre-GPT3 countries can spin up news agencies on the go that aren't obvious to the people that they are state-sponsored.
State-sponsored cybernetic attacks on other states' infrastructure? We pretend for decades that we don't see them, because... well, they are trivial to deny, you just need to play dumb.
My point is that can we actually see them and act against them, even if we publicly deny them. Imagine a social attack that you don't detect but changes how people vote over the decade towards unproductive politicians.
If you believe AGI can come up with new attack vectors (maybe on the net, it might be new bio-weapons), then the defending country needs to spend expenditure on how to detect them. If the defending country can't spin up AGIs of its own or lags behind (imagine North Korea), why wouldn't they just say we can't possibly hope to keep up or defend against an AGI as time progresses. We should nuke AGI installations while our nukes would still do damage.
Please let me know if something doesn't make sense or is just not new information to you.