Registration will open in a few weeks, after we have a chance to review the paper submissions.

We invite industry practitioners, regulators, an non-academic experts on AI accountability and related topics to join us at Wharton on February 6 for valuable interdisplinary conversations.

Also, if Character AI is taking the position that every word generated by the chatbot is actually the company speaking, vs. just indirect output of a tool which they designed, that might open up greater responsibility for firms.

Not to mention that libel, slander, and defamation involve speech by definition!

The Chinese ride-hailing giant Didi has been offering female riders the option of female drivers around the world for soem time.

No doubt. Yet if one accepts the imagined construct that some models explicitly impose their own preferences, vs. striving for some version of fidelity to reality, Grok is the clear proven example. But of course, not the droids they are looking for.

Only one frontier AI model, Grok, has been documented to engage in "top-down ideological bias" prohibited under the US AI Action Plan, by literally checking that certain answers agree with xAI's CEO. Presumably its $200 million DOD contract is about to be revoked?

Given that the government procures the same foundation models that are marketed the private sector, how can it not?

That was a given. Fortunately, if the primary goal is to build AI infrastructure, the laws of economics point toward the cheapest energy sources. Which are, increasingly, renewables.

It will be interesting to see how the FCC implements and legally justifies is role in reviewing state AI legislation. I unerstand the desire to use whatever tools are available to cut red tape for AI development, but the FCC path worries me.