Skip to content

Details

This is going to be an online meetup using Zoom. If you've never used Zoom before, don't worry — it's easy to use and free to join.

Click on the link above at the scheduled date/time to log in...

***

***

HOW A.I. IS REVOLUTIONIZING WARFARE:
HOW CLOSE ARE WE TO SCI-FI SCENARIOS & HOW CONCERNED SHOULD WE BE?

INTRODUCTION:

Artificial Intelligence is no longer only a topic for science fiction writers, philosophers and software engineers. It's becoming part of how states fight wars: identifying targets, coordinating drones, accelerating cyber attacks, and potentially shaping nuclear decision-making. That makes military application of AI an especially useful entry point for our later discussion of AI risk & regulation (scheduled for Sunday, May 31), because it forces us to ask a hard question: when the combat advantages of AI pushes the world's militaries to disregard Isaac Asimov's First Law of Robotics (i.e. "a robot cannot harm a human or allow a human to come to harm through inaction") and train intelligent machines to purposefully harm & kill humans, what does this bode for the future?

The 1st section will look at the rise of semi- and fully autonomous military drones, also known as "lethal autonomous weapons" (LAW). The war in Ukraine has shown how quickly inexpensive drones can reshape the battlefield, and recent reports suggest that AI-assisted drones may now be able to coordinate attacks, resist jamming, and complete missions with less direct human control. We'll discuss how the 2017 film short “Slaughterbots” raised this question in a more concrete form: are autonomous weapons a manageable extension of existing military technology, or do they create a new danger by making killing cheaper, faster, more scalable, and easier to delegate?

The 2nd section will analyze how AI is being integrated into the military's “kill chain”, i.e. the process of analyzing reconnaissance data, finding targets, recommending the type of strikes, approving attacks, and assessing damage. Recent reporting on AI-enabled targeting systems such as Palantir's Project Maven used in the Iran War or Israel's Hasbora & Lavender used in Gaza, suggest that the most consequential AI decisions may happen before anyone pulls the trigger. The central issue here is whether human commanders remain meaningfully in control when AI systems are filtering the battlefield, generating target packages, and compressing what once took several days' worth of analysis into minutes.

The 3rd section examines AI and cyberwarfare. Recent reports about Chinese AI-orchestrated cyber-espionage campaigns using Claude, alongside claims that cyber operations knocked out power to Caracas during the US raid in Venezuela that captured Nicolas Maduro, show how AI can turbo-charge cyber operations. Just as with the military's targeting of enemy sites for bombing & missile strike, AI can help speed up the "cyber kill chain" by automating reconnaissance, identifying weaknesses, selecting malware, finding exploits, and creating more effective phishing operations, but it can also help defenders detect intrusions faster. At the same time, cyber operations can blind radars, disrupt communications, knock out critical infrastructure, and otherwise shape the physical battlefield before kinetic attacks begin.

The 4th section will address the most dangerous possible application: integrating AI into nuclear command & control (NC2) systems. Recent AI nuclear-wargaming studies have found that frontier models may escalate rapidly in simulated crises, while analysts warn that these results reveal both the dangers of machine reasoning and the limits of using AI to model human nuclear decision-making. This section asks where, if at all, AI belongs in nuclear systems—and whether speed & automation are precisely what we should avoid in the one domain where mistakes could be civilization-ending.

RELEVANT MATERIAL FROM PAST MEETUPS:

Back in Mar. 2022, we had a meetup entitled "Are We Headed For World War 3?" The most relevant sections for today's discussion are the 3rd section where we discussed nuclear proliferation & nuclear deterrence theory and the 4th section where we discussed cyberwarfare and the possibility of a "cyber world war" in which critical infrastructure like could be targeted on a massive scale.

DIRECTIONS ON HOW TO PREPARE FOR OUR DISCUSSION:

The videos & articles you see linked below are intended to give you a basic overview of some of the major debates over some of the ways AI is revolutionizing warfare in the 21st century. As usual, I certainly don't expect you to read all the articles prior to attending our discussion. The easiest way to prepare for our discussion is to just watch the numbered videos linked under each section - the videos come to about 50 minutes total. The articles marked with asterisks are just there to supply additional details. You can browse and look at whichever ones you want, but don't worry - we'll cover the stuff you missed in our discussion.

In terms of the discussion format, my general idea is that we'll address the topics in the order presented here. I've listed some questions under each section to stimulate discussion. We'll do our best to address most of them, as well as whatever other questions our members raise. I figure we'll spend about 40 minutes on each section.

***

I. THE RISE OF LETHAL AUTONOMOUS WEAPONS:

  • - Do drones make sense in terms of minimizing the military's casualties, or will they inevitably make war more dangerous in the long run when both sides adopt them, since drone pilots or programmers of autonomous drones have less "skin in the game"?
  • - Are militaries adopting LAWs mainly for speed, cost, survivability against jamming, reduced manpower, precision, or because drone warfare now demands machine-scale coordination?
  • - Is a drone "autonomous" only when it chooses targets and fires, or does autonomy begin earlier—when it navigates, adapts, or continues a mission after being jammed?
  • - Will autonomous drone swarms favor attackers or defenders? Do swarms overwhelm defenses through speed, numbers, and coordination, or will jamming, nets, directed energy, counter-swarms, and layered defenses keep them manageable?
  • - Does the "Slaughterbots" video exaggerate the threat of terrorists or bad state actors using swarms of small autonomous drones to attack civilians, or is this more a matter of when it will happen than if it will happen?
  • - Can states regulate or ban autonomous drones to prevent an arms race? What red lines should democracies set on LAWs?

1a) BFBS Force News, "AI-controlled drone swarms arms race to dominate the near-future battlefield" (video - 4:38 min)
https://www.youtube.com/watch?v=h2O17B4R7Rc

1b) DUST, "Sci-Fi Short Film - Slaughterbots” (video - 7:58 min)
https://www.youtube.com/watch?v=O-2tpwW0kmU

II. THE INTEGRATION OF A.I. INTO THE "KILL CHAIN" (RECON DATA ANALYSIS & TARGET SELECTION):

  • - Does AI merely accelerate the military kill chain, or can it improve judgment and reduce friendly fire and collateral damage?
  • - What should count as meaningful human control of the kill chain - merely authorizing strikes in AI-selected targets, or double-checking target classifications, assessing unintended consequences, etc.?
  • - Does AI make intelligence more reliable by fusing more sources, or can it make old/mistaken information look more authoritative than it really is?
  • - If commanders remain legally responsible for strikes, but AI systems increasingly shape what they see and how fast they decide, do current rules provide enough guidance?
  • - Does Iran's "mosaic defense" suggest that less powerful states can withstand the huge onslaught of AI-enabled strikes in the opening phase of a war? Does this come at the cost of centralized command & control, making miscalculations more common & deescalation harder?

2.) Johnny Harris w/ Shashank Joshi, "This Is How AI Is Rewriting the Rules of War" (video - 18:48 min, start at 4:26)
https://www.youtube.com/watch?v=geaXM1EwZlg&t=4m26s

III. THE INTEGRATION OF A.I. IN CYBERWARFARE:

  • - Why hasn't there been a "cyber Pearl Harbor" or "cyber 9/11" (i.e. cyberattack by a state or non-state actor on critical infrastructure that causes mass casualties) yet, despite two decades' of warning it was imminent? Does AI-enabled cyberattacks make this more likely?
  • - Does AI make cyberattacks more effective (i.e. harder to detect & more damaging), or just faster and cheaper? And is AI-enabled cybersecurity keeping pace with this threat?
  • - Who benefits most from AI-enabled cyber tools: elite state-sponsored hackers or lower-skilled actors like cybercriminals, terrorists, hacktivists, etc.?
  • - If prompt-level guardrails are inadequate because attackers can split a malicious operation into many innocent-looking subtasks, what can be done to prevent misuse of powerful & widely avaliable AI systems like Claude, ChatGPT, Deep Seek, etc?
  • - How much permanent damage can a cyberattack do to critical infrastructure, and how does it compare to a kinetic attack (bomb, missile, etc.)?
  • - How should international law treat cyberattacks on dual-use civilian infrastructure (e.g. power plants, transit systems) - i.e. how should the criteria of proportionality and distinction apply?

3a) Nate B. Jones, "Inside Anthropic's Detection of an AI-Run Cyberattack on 30 High Value Global Targets" (video - 9:57 min.)
https://www.youtube.com/watch?v=7Kc9BNEe2mk

3b) WION, "US Cyber Warfare Powers Venezuela Action, Blinds Radars and Command Networks" (video - 2:18 min.)
https://www.youtube.com/watch?v=MhurhEI8HnE

IV. POTENTIAL A.I. INTEGRATION IN NUCLEAR COMMAND & CONTROL:

  • - Where, if at all, can AI be acceptably integrated into nuclear weapon command & control systems? Is it safe for lower-risk uses like maintenance, logistics, authentication, and wargame support? How about from early-warning assessment, or real-time targeting?
  • - Could AI help leaders process overwhelming data more accurately, or would hallucinations, automation bias, opaque reasoning, bad training data, and cyber manipulation make nuclear crises more dangerous?
  • - Are humans safer decision-makers because they can fear nuclear war, as seen with Cold War near-misses like the Cuban Missile Crisis, Petrov’s 1983 false alarm, and Able Archer?
  • - Do wargames show that LLMs are dangerously risk-prone when it comes to using nuclear weapons, or do they merelt reveal that LLM models are trained on nuclear-strategy literature biased toward deterrence and escalation?
  • - If AI improves attacks on satellites or conventional strikes against the nuclear triad (ICBMs, bombers, subs), could it undermine second-strike stability and thus trigger nuclear escalation without ever “controlling” nuclear weapons?
  • - Can international agreements meaningfully limit states from using AI in nuclear command & control, considering that this is much harder to verify than nuclear enrichment, nuclear tests or missile silos?

4a) CBS News, "How AI could be incorporated into nuclear weapons" (video - 1:55 min)
https://www.youtube.com/watch?v=kNpzMy2ZHuQ

4b) Reid Hoffman & Aria Finger, "Why AI Chose The Nuclear Option" (video - 33:01 min, listen to 9:30)
https://www.youtube.com/watch?v=Y5DBF9Fh1gg

XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX

Related topics

Artificial Intelligence
Military
AI Ethics
Autonomous Drones
Foreign Policy Topics

You may also like