Schedule

Join us to discuss what we should mean by "Slow down"/"Pause AI”/"Pace the frontier"?

Do they all basically mean the same thing? Does it even matter, so long as we do something? Should we pick whatever gets the most consensus? Pursue what's most optimal, and reject half measures?

Different proposals coalesce toward different goals:

Dario’s “We Must Pace the Frontier” calls for slower capability gains while training continues, potentially buying an extra year or two for safety work.
July’s AI-worker letter, now signed by 1,386 AI lab employees, asks governments to build the international tools needed to make slowing down possible, without demanding an immediate halt or specifying its duration.
AI 2040’s Plan A combines controlled progress with a near-pause at roughly top-human-expert capability from 2035 to 2040.
PauseAI US seeks a global treaty banning superintelligent AI until it can be made safe.
PauseAI International would halt training the most powerful general AI systems until safety and democratic control are secured.
The Statement on Superintellgence, calls for a prohibition that lasts until there is broad scientific agreement on safe, controllable development and strong public buy-in.
Eliezer Yudkowsky’s “Shut it all down” calls for an indefinite, worldwide moratorium on large AI training runs..
Sanders and Casar’s proposal would permanently ban superintelligence while temporarily pausing other advanced AI development until a federal regulator establishes safety rules.
Katja Grace’s “Let’s think about slowing down AI” suggests slower progress, redirecting research, and halting categories of work until safe.

Which of these should we want to get behind? How much do these differences matter, if at all? Is something better than nothing?

We’ll discuss which differences are load-bearing, how to avoid letting the perfect be the enemy of the good, and whether a shared call to “pause AI” builds momentum or obscures incompatible goals.

Join us to discuss what we should mean by "Slow down"/"Pause AI”/"Pace the frontier"?

Do they all basically mean the same thing? Does it even matter, so long as we do something? Should we pick whatever gets the most consensus? Pursue what's most optimal, and reject half measures?

Different proposals coalesce toward different goals:

Dario’s “We Must Pace the Frontier” calls for slower capability gains while training continues, potentially buying an extra year or two for safety work.
July’s AI-worker letter, now signed by 1,386 AI lab employees, asks governments to build the international tools needed to make slowing down possible, without demanding an immediate halt or specifying its duration.
AI 2040’s Plan A combines controlled progress with a near-pause at roughly top-human-expert capability from 2035 to 2040.
PauseAI US seeks a global treaty banning superintelligent AI until it can be made safe.
PauseAI International would halt training the most powerful general AI systems until safety and democratic control are secured.
The Statement on Superintellgence, calls for a prohibition that lasts until there is broad scientific agreement on safe, controllable development and strong public buy-in.
Eliezer Yudkowsky’s “Shut it all down” calls for an indefinite, worldwide moratorium on large AI training runs..
Sanders and Casar’s proposal would permanently ban superintelligence while temporarily pausing other advanced AI development until a federal regulator establishes safety rules.
Katja Grace’s “Let’s think about slowing down AI” suggests slower progress, redirecting research, and halting categories of work until safe.

Which of these should we want to get behind? How much do these differences matter, if at all? Is something better than nothing?

What should we mean by "Slow down"/"Pause AI”/"Pace the frontier"? (Group Discussion)

What should we mean by "Slow down"/"Pause AI”/"Pace the frontier"? (Group Discussion)