Amdocs

We need a plan to govern the development of AI, we need global buy-in on guardrails that should be implemented to prevent the worst possible outcomes. I don't expect the first meeting to solve everything, but I think if everyone can agree we need to do something - that would be better for humanity than doing nothing.

• Safe, secure and trustworthy AI
• Protection and promotion of human rights
• Transparency, accountability, and human oversight
• Social, economic, ethical, cultural, linguistic and technical implications of AI

Some governments have chosen not to regulate, to enable companies to develop faster, but we are running toward a cliff. How far are we from AGI? I do not know, but if we don't get guardrails in place to protect human rights we will witness atrocities we cannot imagine.

I think we need to consider the implications on capitalism, or more to the point - government funding. Even outside of capitalist economies, the money governments rely on will be impacted. Big business, will be impacted. I'm not sure that angle receives enough attention under the general heading of "economic ... implications of AI"

In the US, we've chosen not to regulate and as a result, Open AI, Anthropic, and others, seem to have largely abandoned their safety initiatives.

Leaders need more education on AI, there are so few Technologists in seats of power, I do not believe they have the foundational knowledge required to build the necessary regulatory guardrails. We need to educate!

This can't just be a technology-led intiaitive, we need technologists, ethicists, cultrual anthropologists, psychologists, medical doctors, economists, lawers... from many countries... we need to look at this from all angles. Right now, it's a handful of men in America running the whole thing.

The advent of LLMs is what shifted the development of AI into high gear. But, have we considered how limiting using language actually is? Linguists can help us explore this. When we consider the demographics of the authors of text included in the corpus for training these models, we have disproportionate representation of social classes and groups; how does that impact bias? Perhaps it would be helpful to center discussions on a single topic so people can understand the cross-functional/disciplicary effects.

Minorities, whether that be racial, social, economic... basically anyone who isn't a straight, white, male, is underrepresented.

Lock them in a room and don't let them out until they gain understanding of each other's perspectives. I'm only slightly joking...

You can't govern what you don't understand. I think efforts, especially for this first meeting, needs to focus on building the foundational knowledge required to explore future topics. Everyone needs to be on the same page first.