We pronounce disagreement with “Pause AI” movement, which calls on possibly government-enforced immediate stop of “powerful AI training” for at least 6 months:
https://futureoflife.org/open-letter/pause-giant-ai-experiments/
Calls to stop progress with any new technologies aren’t anything new. It was there when the first automation in manufacturing appeared. It happened in many industries when there was fast progress leading to change.
The reasons for those calls were the same: people are naturally afraid of change, especially if it’s rapid. Policymakers want control and some established industry players don't like disruption.
Every time there was some catastrophic vision about society, loss of jobs, machines will replace humans, etc.
Fortunately those calls weren’t much successful, those visions weren’t correct, technological progress makes our society more wealthy as a whole and it enables humankind to thrive.
This call to “pause progress” isn’t anything new either. Of course it claims “this time it’s different”. Today’s innovation will again automate away all the jobs. It will lead again to catastrophic outcomes for society on many levels.
Maybe it is a bit new in how exaggerated some statements are. Claim that we “lose control of our civilization”, due to large language models sounds quite funny, when most experts agree those models aren’t close to general artificial intelligence.
Large language models will augment human abilities, enable humans to achieve more, and make society more prosperous in the end – exactly like any other new important technology. That's great and we really shouldn't pause it just because some people think so and some politicians eventually take it as their marketing topic.
Therefore we are saying no to anything like “government step in to institute moratorium” on progress in this area. It’s a ridiculous, unprecedented, and net negative for the human future.
Apart from technological progress arguments, there are also strong practical implementations and geo-political considerations.
Is anything like an arbitrary “worldwide pause” on technological progress in this area implementable? Would the majority of countries agree to that, and if so, would they really comply?
Who would benefit if only Western countries such as US and Europe comply? Wouldn’t they lose their leading position or at least competitiveness? Do we want this outcome?
Honest answers to these questions lead us to an even stronger conclusion that anything like “Pause AI” initiative should be strongly discouraged.