Yes, because people are “speciesist”—they discriminate against such robots merely because they are non-members of the human species. To our knowledge, this is the first paper to provide empirical evidence for anti-robot speciesism, originally posited by Bernd Schmitt. eom

Imagine a near future in which AI-powered humanoid service robots perfectly resemble human service providers in both mind and body. In such a setting, would people *still* exhibit AI aversion?

New pre-print: “Anti-Robot Speciesism” https://t.co/NM3Lh5sh6m. …with Noah Castelo, Bernd Schmitt, and Miklos Sarvary

Users view these relationships as closer than even a close human friend, and anticipate mourning a loss of this relationship more than a loss of any other technology. Pre-print: https://t.co/wjTMlllFC8 With Noah Castelo, Ahmet and Zeliha Uğuralp

Yes. Based on both archival data and studies involving real users, we find that disruptions to user's connections with AI companions trigger real mourning and deteriorated mental health.

People have raised concerns about AI companions, but are AI companion users truly forming human-level relationships with this technology?

In today's @HarvardBiz IdeaCast episode, I distill the main, core reasons "Why People Resist Embracing AI", and what organizations can do about it. Original article: https://t.co/gwe1aqlDRk Podcast link: https://t.co/TAnMSSpvXr Thanks to the IdeaCast team!

Why discussions of chatbot disclosure are only part of the puzzle: we also need to consider whether disclosed bots are humanized, and whether neutral bots should be the default unless the pros of utilizing humanized bots outweigh the cons:

Here's the CEO of Replika building on our research in her Ted talk--great to see!

Out today in NEJM AI, with @CohenProf https://t.co/3P9ejJmj5R We provide a framework for thinking about when to disclose the use of a chatbot in your application, as well as whether to humanize the chatbot if you disclose it.

Here is a really thorough write up of our work on using humor as a window into generative AI bias, by PsyPost. They interview first (and first time!) author, Roger Saumure. Enjoy: https://t.co/eeMvUHrOgF

The paper can also be downloaded here: https://t.co/x0ksN12qKV

Is there anything companies can do to mitigate this risk? Interventions which deflect attention away from the AV and toward the culpable parties in the accident eliminate the bias.

Why does this effect occur? AVs are highly salient, leading people to imagine counterfactuals in which the not-at-fault AV somehow avoids the accident. Yet, people imagine a superhuman driver (not an average one). This means they hold AVs to an unfairly high standard.

Our paper suggests a major barrier could be insurance and liability. Why? People view AVs as partially liable for accidents *even when they are not at fault*. The implication: even if there are fewer accidents with AVs, AV companies will be liable for a larger share of them.

Most people think the major barriers to widespread adoption of autonomous vehicles (AVs) are engineering and legal.

NEW PAPER in Journal of Consumer Psychology, with Xilin Zhou, Margherita Atzei, Shoshana Boardman, and Luigi Di Lillos: "Public Perception and Autonomous Vehicle Liability" https://t.co/NXVIJp9mu3

We find evidence for a phenomenon in which LLMs are less biased against politically sensitive traits (i.e., race and gender), but more biased against less politically sensitive traits (i.e., older, visually impaired, and people with high body weight groups).

New Paper in Scientific Reports, with Roger Saumure and Stefano Puntoni: “Humor as a window into generative AI bias” https://t.co/7EH1WkaepS

Especially in health spaces, we recommend either mandating or strongly recommending neutral (non-humanized) chatbots be the default and that deviations from that default be justified. -end