This is an automated rejection. No LLM generated, assisted/co-written, or edited work.
Read full explanation
As I was going home from the Champaign airport after an eventful EAG conference, I struck up a conversation about AI safety with my Lyft driver.
I thought to myself, “It’d be good to see what the public thinks. Maybe inform him of the risks. It’s marginal but positive counterfactual.”
But as I spoke to him, he started dropping words like “UBI” and drawing analogies to power concentration and gradual disempowerment. I was surprised to see how much he understood already.
The conversation went one way or another, and he started talking about his experience in the military.
He spoke about the efficiency of the military, how they function with clear hierarchies, well-defined responsibilities, and straightforward instructions. Amidst his thorough and proud explanation, I began to envy the clarity and discipline of the system he was describing.
He kept talking about how life in the military is completely structured. How instructions are straightforward and people know what to do, and make sure to do it.
I said to him, “So none of your work is bottlenecked by other people.”
He replied:
“If we have bottlenecks, people die.”
That stuck with me.
Isn’t this the same situation we are in?
If we genuinely believe there's a non-negligible chance of catastrophic outcomes from AI on short timelines, shouldn't that be reflected in how seriously we approach execution?
I'm not suggesting AI safety should adopt military hierarchies. Independent thinking, disagreement, and experimentation are essential to the field.
But perhaps there's something to learn from the mindset.
In the military, the possibility of lives being lost isn't just something people acknowledge intellectually. It shapes how they train, prepare, make decisions, and carry out their responsibilities.
I wonder how much of that same seriousness we've internalized in AI safety.
It's one thing to believe that the stakes are existential. It's another to let that belief meaningfully change how we work, what we prioritize, and what standards we hold ourselves to.
If we genuinely believe what's at stake, shouldn't we act like it?
As I was going home from the Champaign airport after an eventful EAG conference, I struck up a conversation about AI safety with my Lyft driver.
I thought to myself, “It’d be good to see what the public thinks. Maybe inform him of the risks. It’s marginal but positive counterfactual.”
But as I spoke to him, he started dropping words like “UBI” and drawing analogies to power concentration and gradual disempowerment. I was surprised to see how much he understood already.
The conversation went one way or another, and he started talking about his experience in the military.
He spoke about the efficiency of the military, how they function with clear hierarchies, well-defined responsibilities, and straightforward instructions. Amidst his thorough and proud explanation, I began to envy the clarity and discipline of the system he was describing.
He kept talking about how life in the military is completely structured. How instructions are straightforward and people know what to do, and make sure to do it.
I said to him, “So none of your work is bottlenecked by other people.”
He replied:
“If we have bottlenecks, people die.”
That stuck with me.
Isn’t this the same situation we are in?
If we genuinely believe there's a non-negligible chance of catastrophic outcomes from AI on short timelines, shouldn't that be reflected in how seriously we approach execution?
I'm not suggesting AI safety should adopt military hierarchies. Independent thinking, disagreement, and experimentation are essential to the field.
But perhaps there's something to learn from the mindset.
In the military, the possibility of lives being lost isn't just something people acknowledge intellectually. It shapes how they train, prepare, make decisions, and carry out their responsibilities.
I wonder how much of that same seriousness we've internalized in AI safety.
It's one thing to believe that the stakes are existential. It's another to let that belief meaningfully change how we work, what we prioritize, and what standards we hold ourselves to.
If we genuinely believe what's at stake, shouldn't we act like it?