Debate Dossier
AI Policy · Live Motion
Should AI Be Banned?
A maximalist AI debate motion. The Pro burden is heavy; the Con burden is to show the benefit is not replaceable.
FormatQuick Clash / BP / PF adaptable
DifficultyHard
Main clashRisk avoidance vs benefit forfeiture
Best forCounterfactual reasoning, Burden allocation, Tight motions
The round turns on this
Is the harm only avoidable by prohibition?
Ban
- Harm is structural, not patchable
- Enforcement is possible at compute layer
- A pause is not a permanent ban
Do not ban
- Benefits are not replicable by other tools
- Ban displaces, does not stop
- Tiered rules already cover the cases
Pro must show prohibition is the only fix.
Argument arena · prep both sides
Pro
When harm is structural and present-tense, the only honest answer is to stop building the system that produces it.
PRO 1 Structural harm
ClaimBias, displacement, and capture are features of the technology, not bugs.
WarrantPatches address symptoms; the underlying scaling dynamic continues.
ImpactNo targeted rule fixes a structural problem.
Attack this
Con will say structural problems still have non-prohibition fixes.
PRO 2 Compute is a chokepoint
ClaimFrontier AI runs on a small number of identifiable chips and data centers.
WarrantUnlike speech or research, the substrate is physical and licensed.
ImpactEnforcement is feasible in a way that maximalist bans usually are not.
Attack this
Con will say compute governance is not the same as a ban.
VS
Con
A blanket ban surrenders real benefits and displaces, rather than eliminates, the harms.
CON 1 Benefit forfeiture
ClaimAI is already the best system at protein folding, fraud detection, and accessibility.
WarrantThese benefits are not replicable by older tools at the same cost.
ImpactYou lose the upside while the misuse migrates.
Attack this
Pro will say "non-frontier" carve-outs are still possible.
CON 2 Displacement, not elimination
ClaimA national ban does not stop development globally.
WarrantAdversaries keep building and inherit the standard-setting power.
ImpactYou buy zero safety and lose your seat at the table.
Attack this
Pro will say coordinated bans on dangerous capabilities have worked before (CFCs, bioweapons).
Sample round · flowed with judge notes
Pro · openingStrong open
The harms are structural and the substrate is physical. Compute governance makes a prohibition enforceable in a way speech bans never were.
JudgeStrong mechanism. Heavy burden remains.
Con · responseBest turn
You forfeit the upside and you displace the downside. Protein folding does not stop globally because Washington stops domestically.
JudgeClean counterfactual. Pro must answer migration.
Pro · rebuttalAnalogy
Bans on bioweapons and CFCs worked because the harm crossed a threshold worth coordinating on. AI fits that pattern.
JudgeStrong analogy. Con can dispute the threshold.
Con · weighingBurden
The motion is "ban AI," not "ban frontier autonomous systems." On the motion as worded, the burden is impossible. Take Pro's narrower fix and tiered regulation does it.
JudgeBurden play wins the room.
Judge ballot
Con wins
Wide margin
Reason for decision
Pro's strongest moments are about a narrower motion than the one debated. On "ban AI" as worded, Con's benefit-forfeiture and migration arguments are unanswered.
Key clash
Is the motion "ban AI" or "ban frontier autonomous AI."
Pro · feedback
Concede the breadth of the motion early and pick a narrower mechanism. You needed risk-tiering, not prohibition.
Con · feedback
Excellent burden frame. The analogies to CFCs deserved a harder response.
One drill before the rematch
Run Pro on a narrower motion ("This house would ban autonomous AI weapons") and see if the case lands.