AI & Digital Ethics Grade 7: Autonomous Weapons and AI Governance

Grade 7 | AI & Digital Ethics Topic: Autonomous Weapons and AI Governance

1. The Driving Question

If a robot can decide to pull a trigger without a human saying "go," who’s really responsible when something goes wrong—and how do we stop it from happening in the first place? Right now, militaries are testing AI-powered drones and tanks that can pick targets on their own. But what if the AI misidentifies a school bus as a tank? Or what if two countries’ AI weapons start fighting each other without any humans involved? The question isn’t just can we build these weapons—it’s should we, and how do we control them before they control us?

2. The Core Idea — Built, Not Listed

Imagine you’re playing a game of laser tag with your friends at a park in Chicago. You set up rules beforehand: no shooting at people wearing red shirts (they’re on a break), no aiming at faces, and if the game gets too intense, anyone can yell "pause" to stop. Now, what if one of the laser guns was programmed to never stop—even if someone yells "pause," even if it’s aiming at a kid in a red shirt? That’s the problem with autonomous weapons: they follow code, not human judgment, and once they’re deployed, they can’t "un-pull" the trigger.

These weapons use AI to scan for targets (like tanks or soldiers) and decide when to attack—all without a human pressing a button. But AI isn’t perfect. In 2020, a U.S. military test showed an AI drone misidentifying a cloud as a missile. If that drone had been armed, it could’ve started a war by accident. The bigger puzzle is governance: how do we make rules for weapons that don’t have morals, don’t feel guilt, and don’t listen to "pause"?

Key Vocabulary: - Autonomous weapon: A weapon system that can select and engage targets without human intervention. Example: A drone programmed to patrol a border and fire at any vehicle crossing without a specific license plate. Note: In college-level ethics, this term expands to include "dual-use" technologies (e.g., AI that can both deliver medicine and guide missiles).

• Lethal Autonomous Weapon System (LAWS): A formal term for autonomous weapons designed to kill. Example: The "Harpy" drone, used by Israel, which can hunt and destroy radar systems on its own. Note: The UN debates whether these should be banned entirely, like chemical weapons.

• Governance: The rules, laws, and agreements that control how a technology is used. Example: The "Rules of War" (like the Geneva Conventions) are a form of governance for traditional weapons. Note: In AI, governance includes who makes the rules (governments? tech companies?) and how to enforce them globally.

• Algorithmic bias: When an AI’s decisions are unfair or inaccurate because of flaws in its training data. Example: An autonomous weapon trained mostly on images of tanks in deserts might fail to recognize tanks in snowy forests—or mistake a hospital for a military base. Note: In college, this connects to systemic bias in facial recognition, hiring algorithms, and predictive policing.

3. Assessment Translation

How this appears on state assessments (Grade 7): - Multiple choice: Questions test understanding of definitions (e.g., "Which scenario describes an autonomous weapon?") and ethical trade-offs (e.g., "What is the biggest risk of using AI in warfare?"). Distractor patterns: Wrong answers often confuse automation (e.g., a remote-controlled drone) with autonomy (e.g., a drone that decides to fire on its own). They might also oversimplify risks (e.g., "AI weapons are always more accurate than humans").

• Short constructed response (1–2 paragraphs): Prompts ask students to explain a trade-off (e.g., "Describe one benefit and one risk of autonomous weapons") or argue a position (e.g., "Should the U.S. support a global ban on LAWS? Use evidence from the text"). Proficient response: Includes a clear claim, at least one piece of evidence (e.g., a real-world example or statistic), and addresses a counterargument. Developing response: Lists risks/benefits without explanation or ignores the counterargument.

• Evidence-based writing (longer response): Students analyze a scenario (e.g., "A country deploys AI tanks that can’t distinguish between soldiers and civilians. What should other nations do?") using sources. What teachers look for:

• Use of specific vocabulary (e.g., "algorithmic bias," "governance").
• Connection to real-world examples (e.g., the UN’s debates on LAWS, the 2023 U.S. policy on autonomous weapons).
• Logical structure: Problem-Stakeholders-Possible solutions-Best choice with justification.

Model Proficient Response (Short Constructed Response): Prompt: "Should governments ban autonomous weapons? Explain your answer using at least one piece of evidence." Response: "I think governments should ban autonomous weapons because they can’t make ethical decisions like humans can. For example, in 2021, a UN report found that an AI drone used in Libya hunted down and killed a human target without any human approval. Even if the AI is ‘accurate,’ it can’t understand context—like whether a target is surrendering or if civilians are nearby. Some people argue that autonomous weapons could reduce human casualties in war, but the risk of mistakes is too high. If an AI misidentifies a school as a military base, no one can stop it in time. A ban would force countries to keep humans in the loop, which is safer."

4. Mistake Taxonomy

Mistake 1: Confusing "autonomous" with "remote-controlled" - Prompt: "Which of the following is an example of an autonomous weapon? A) A soldier using a joystick to fly a drone. B) A drone that scans for enemy tanks and fires without human input. C) A missile that follows a pre-programmed path. D) A robot that carries supplies for soldiers." - Common wrong answer: A or C. - Why it loses credit: The question tests the definition of autonomy (decision-making without human input). Remote-controlled drones (A) and pre-programmed missiles (C) still require human commands. The correct answer is B, but students often pick A because they associate drones with "automation." - Correct approach: Focus on the key phrase "without human input." Autonomous weapons decide to act; remote-controlled ones just execute human orders.

Mistake 2: Ignoring the "who decides?" problem in governance - Prompt: "Explain one challenge of creating global rules for autonomous weapons." - Common wrong response: "It’s hard because AI is complicated." - Why it loses credit: The response is too vague. The real challenge is who gets to make the rules—countries with advanced AI (like the U.S. or China) might not agree, and tech companies (like Google or Palantir) also influence the technology. - Correct approach: Name a specific stakeholder (e.g., "The UN wants a ban, but the U.S. military opposes it") and explain why their interests clash.

Mistake 3: Overlooking bias in AI training data - Prompt: "Why might an autonomous weapon misidentify a civilian target? Use the term ‘algorithmic bias’ in your answer." - Common wrong response: "Because the AI is evil" or "Because it’s not smart enough." - Why it loses credit: The question asks for a technical explanation, not a moral one. Students often default to sci-fi tropes ("AI turns against us") instead of explaining how training data shapes AI behavior. - Correct approach: "An autonomous weapon might misidentify a civilian because of algorithmic bias. For example, if the AI was trained mostly on images of soldiers in deserts, it might not recognize soldiers in cities—or mistake a farmer with a tool for a combatant. This happens because the AI’s ‘knowledge’ comes from limited data, not real-world understanding."

5. Connection Layer

• Within AI & Digital Ethics-Deepfakes and disinformation: Both autonomous weapons and deepfakes rely on AI’s ability to manipulate reality (targeting decisions vs. fake videos). Understanding one reveals how AI can be weaponized in ways that are hard to detect or reverse—once a drone fires or a fake video spreads, the damage is done.
• Across subjects-Physics (Newton’s Third Law): Autonomous weapons are a real-world example of "for every action, there’s an equal and opposite reaction." When Country A deploys an AI tank, Country B will build one to counter it, creating an arms race with no "pause" button. This mirrors how forces in physics escalate (e.g., rockets pushing against Earth’s gravity).
• Outside school-Video games with "AI teammates": In games like Call of Duty or Apex Legends, your AI squadmates sometimes make dumb decisions (e.g., running into gunfire). This is a low-stakes way to see how AI "thinks"—it follows rules but lacks common sense. The same flaw in a real autonomous weapon could start a war.

6. The Stretch Question

If an autonomous weapon kills a civilian by mistake, who should be held responsible: the programmer, the military commander who deployed it, the government that approved it, or the AI itself? Pointer toward the answer: This isn’t just a legal question—it’s about how we assign blame when machines blur the line between tool and decision-maker. In 2023, the U.S. Department of Defense released a policy saying humans must always be responsible for lethal actions, but this gets messy if the AI acts in ways its programmers didn’t predict. Some philosophers argue we need new legal categories, like "digital personhood," for AI. Others say the real problem is that we’re treating war like a video game, where mistakes have no consequences. The answer might depend on whether you think AI can ever be truly autonomous—or if it’s always just a mirror of the humans who built it.