Artificial Intelligence · Real arguments from SuperDebate members below
AI capability gains have repeatedly surprised the experts building these systems. GPT-4 passed the bar exam, aced medical licensing tests, and wrote publishable academic papers — none of which was predicted one year before it happened. We cannot assume alignment will be solved in time when the...
Posted by jconnor
A sufficiently advanced system optimizing for a misspecified goal could pursue that goal in ways catastrophically harmful to humans. The paperclip maximizer is a toy example, but the underlying logic — that an optimizer doesn't intrinsically value human welfare — applies to any system whose...
Posted by jconnor
Getting alignment wrong even once could be irreversible. Unlike other technological risks — pandemics, nuclear weapons, climate change — a misaligned superintelligent system may have both the capability and the motivation to prevent correction. The asymmetry demands treating this seriously, not...
Posted by jconnor
There is no credible path from current LLMs to a system capable of outmaneuvering human civilization. Current models are pattern-matching engines that can't reliably count to ten without error. The gap between stochastic text prediction and autonomous world-altering agency is not an engineering...
Posted by jconnor
Existential risk framing distracts from concrete AI harms happening today: algorithmic discrimination in hiring, credit, and criminal justice; deepfake disinformation at scale; and surveillance infrastructure sold to authoritarian governments. These harms are measurable, affect real people now, and...
Posted by jconnor
Every major technology has been accompanied by apocalyptic predictions that did not materialize. Nuclear power would melt cities; the internet would destroy social fabric; GMOs would cause ecological collapse. The track record of catastrophism as a policy guide is poor. The answer to powerful...
Posted by jconnor
Debate this exact resolution against the AI coach and get scored feedback.
Practice with the coachAI capability gains have repeatedly surprised the experts building these systems. GPT-4 passed the bar exam, aced medical licensing tests, and wrote publishable academic papers — none of which was predicted one year before it happened. We cannot assume alignment will be solved in time when the... (Argued by jconnor on SuperDebate.)
There is no credible path from current LLMs to a system capable of outmaneuvering human civilization. Current models are pattern-matching engines that can't reliably count to ten without error. The gap between stochastic text prediction and autonomous world-altering agency is not an engineering... (Argued by jconnor on SuperDebate.)
Not yet. Anyone can take a side on the topic page and challenge an opponent to argue it live on SuperDebate.
On SuperDebate. Post a written argument on the topic page, rehearse the resolution against the AI debate coach, or take it to a live debate night at a club near you. Joining is free.
