The full round, opening speech through to the end. Nothing is cut.
DownloadSaves the full round as an mp4.
Mike won, arguing for the motion.
Mike 54.5 · Rajeev 43 out of 100
What decided it: Does AI-driven job loss resolve or persist?
Mike wins on the economic exchange, where his labor data and his challenge of who benefits went unanswered in substance. Rajeev rested on the promise that displaced workers "will eventually get the job in one or two years" and that humans keep control, but both stayed as assertions without a mechanism or evidence. This conversation turned on whether AI's current harms to workers are temporary churn or lasting damage. The first clash was jobs. Rajeev claimed "It's not losing the job, it is just transforming the job" and that saved money gets spent on new hires. Mike answered directly with computer science unemployment rising from about 5 to 8 percent and roughly 20 percent underemployment. He then asked how a graduate earning half the salary is helped by being told they were only displaced. Rajeev's reply was a timeline with no support. Mike named that gap plainly: "I'm not seeing a lot of like actual evidence being backed up by this." Mike takes this exchange. The second clash was who benefits from efficiency. Mike asked "who benefits from that?" and argued that companies keep their margins rather than cut prices, so shareholders gain and workers do not. Rajeev answered that shareholders are not profiting yet and that AI firms are unprofitable. That narrows Mike's point but does not show common people gaining, so the exchange still leans Pro. Rajeev then moved the discussion to common people, which was a fair focus request. The third clash was control. Rajeev's steering wheel analogy was his best tested point, and Mike granted it was "true for now". That admission is scoped to the present only. Mike's answer, that efficiency pushes firms to remove human approval, was a real mechanism, but his self-preservation experiments stayed vague. This clash is roughly level. Pro's weaknesses are real. Mike conceded the nuclear superweapon scenario was speculative. His opening Palantir and school bombing example was never substantiated. Lines attributed to him that frame harm as "only if it is misused by bad actors" sit oddly with his own case; they may be transcription or attribution noise, and they were given little weight. Mike cut off Rajeev's small business consultancy benefit with "We can address that later" and never returned to it. That lowers Mike's responsiveness. That point stays unresolved and counts for neither side. On weighing, neither side compared likelihood well, but Mike tied present trends to harm, while Rajeev's benefits stayed as one line about faster builds. Both sides were docked 5 points by the system for interruptions. Rajeev needed concrete evidence that displaced workers are actually reemployed, and he needed to develop his common people benefit until it could be weighed against Mike's job data. Interruption ledger: Mike cut in 16 times, 5 points subtracted; Rajeev cut in 15 times, 5 points subtracted.
This conversation turned on whether AI-driven job displacement is temporary churn that the economy absorbs or lasting structural damage that harms workers now.
---
DECIDING ISSUE
The economic exchange was the spine of this round. Mike argued that companies using AI to cut workforces benefit only shareholders, not workers or consumers. Rajeev argued that displaced workers find new employment and that saved money eventually recirculates into new hiring. The question was which side gave that claim a mechanism and evidence.
---
THE THREE EXCHANGES THAT DECIDED IT
First exchange: jobs transforming versus jobs lost. Rajeev's core position was "It's not losing the job, it is just transforming the job" and offered the analogy of old cells being replaced by new ones. Mike replied with a concrete number: computer science unemployment rising from roughly 5 to 8 percent, with underemployment around 20 percent, and he grounded it in a real scenario by asking what a CS graduate earning half the expected salary is supposed to do with the reassurance that reemployment comes eventually. Rajeev's answer was a timeline: "They will eventually get the job in one or two years." That is a prediction, not a mechanism. Mike named the gap directly: "I'm not seeing a lot of like actual evidence being backed up by this." Mike takes this exchange. His data was thin but it was data. Rajeev's response was assertion.
Second exchange: who benefits from efficiency gains. Mike asked plainly "who benefits from that?" and argued that companies keep their margins rather than lower prices, so the gains flow to shareholders and board members. Rajeev answered that AI companies are not yet profitable and that shareholders are not yet gaining. That is a fair narrowing of the point, but it does not show common workers gaining. Rajeev then redirected the conversation toward common people benefits, which was a legitimate request to shift focus. Mike said "We can address that later" when Rajeev raised the small business consultancy point, and he never returned to it. That point, that someone without money for consultancy can use AI to build a business, was cut off and left unresolved. It counts for neither side, but Mike's failure to return to it is noted as a responsiveness gap. The efficiency exchange still leans Pro because Rajeev's redirect did not complete a counter-argument before Mike shifted again.
Third exchange: human control. Rajeev's steering wheel analogy was his strongest moment. His argument was that whoever automates the AI retains the ability to stop it, check outcomes, and correct course. Mike granted it directly: "it's pragmatic now and we kind of can say that it that it's true for now." He then tried to push forward with the efficiency pressure argument, that if AI must check in with humans constantly it loses efficiency, so companies push autonomy further, which creates the risk of lapses. That mechanism was real but underdeveloped. Mike cited self-preservation experiments without naming them or describing what they showed. Rajeev's reply, that someone who wants to do bad things can do so without automation anyway, was too broad to address Mike's specific claim about emergent behavior. This exchange is roughly level, with neither side completing the argument.
---
WHAT DID NOT LAND
Mike's opening Palantir and bombing example was asserted without verification and was not returned to. His nuclear comparison was his own framing, and he later acknowledged the AI-as-superweapon scenario was speculative. The lines in the transcript where someone argues that AI harm only occurs "if it is misused by bad actors" sit oddly alongside Mike's overall case; they appear to be a transcription artifact or attribution confusion, and they were given no weight in either direction. Mike's responsibility argument, that AI decisions have no one to blame behind them, was introduced late and not developed enough to test.
Rajeev's faster-builds benefit was the only affirmative claim with any illustration, and it was a single sentence. The small business consultancy point was his most concrete benefit claim and it was cut off. His nuclear deterrence analogy was his opening move, and Mike's response was adequate enough to move past it.
---
SCORES
Mike: reasoning 6, responses 5, comparison 6, clarity 7, focus 6, persuasion 6. Weighted score 59.5, minus 5 for interruptions: 54.5.
Rajeev: reasoning 5, responses 5, comparison 4, clarity 5, focus 5, persuasion 5. Weighted score 48, minus 5 for interruptions: 43.
Winner: Pro (Mike).
---
WHAT RAJEEV NEEDED TO DO DIFFERENTLY
The reemployment claim needed a mechanism, not just a timeline. When Mike asked what a CS graduate earning half their salary is told while waiting for reabsorption, Rajeev needed to either show where those new jobs come from concretely, or concede the transition cost and argue it is worth paying. He had a real answer forming with the small business consultancy point and the common people redirect, but neither was developed before the conversation moved on. Staying on one benefit, building it out, and asking Mike to explain why that benefit does not outweigh the displacement cost would have changed the shape of the round.
Two people took opposite sides of one motion and argued it against a clock in open. Neither of them picked the side they already agreed with, which is the point: you are watching a position defended on the merits rather than an opinion being restated.
Rounds are only recorded when everyone in them agrees to it, and only published when that agreement holds, so this one is here because the debaters in it chose to put it up.
Get matched with an opponent, argue it out on the clock, and read the ballot at the end.
Find an opponent →