I think we do need to be more focused on safety, protection, and regulatory engagement, given the progress we expect in the field. But I disagree with the view that we may put product before safety. Why do you think they say that? Moderator: Because these are the people you work with? IR RI: I think you need to ask them, but I think many people think that safety and capability are separate. Some people think that safety and capability are separate units, starting with I, I come from the aviation and automotive industries, which have very mature safety thinking and systems, and do not need to constantly debate safety issues because these are already mature.
So I think the entire industry needs to move indonesia phone number details towards a more rigorous safety discipline. We have safety systems, we have strict operational safety discipline, and that includes many aspects, not only operational discipline, but also the safety of our products and deployments today, such as harmful bias, false information, misinformation classifiers, etc. Moderator: I turned off my phone and the notification sound. IR RI: Continue. Not only the alignment of models today, we use human feedback for reinforcement learning, but also think about the long-term alignment of models as they increase in capability, which is a very focused research area.
Moderator: But this has always been in I. I think because you are the current leading company, this problem is more prominent. But people are worried about safety, and one of the purposes of I was founded is to ensure safety. Last week, a group of current and former I and Google employees jointly issued an open letter calling on the company to allow them to warn about the risks of advanced I. This is not new, but in this case, the employees said, "Extensive non-disclosure agreements prevented us from expressing concerns, and the company may have failed to address these issues.