Why do politicians endorse nationalistic policies like vaccine hoarding? In our new paper @PsychScience we test public & expert opinions on vaccine policies, w @mollycrockett JimEverett @SennJulien @MichelAMarechal and in collaboration w @apoliticalco 1/n.

RT @birchlse: 🦀🐝🧠🤖 How should we make decisions when faced with systems of uncertain sentience? My new book The Edge of Sentience: Risk and….

RT @mpshanahan: New postdoc opening at @imperialcollege to work with me on #AI topics, especially at the intersection of AI and philosophy,….

RT @NatureHumBehav: Using a computational model to quantify difficulty in reconstructing images from compressed codes, @qi_lin7 et al. show….

RT @KristinAndrewz: The New York Declaration of Animal Consciousness was signed today by leading figures in philosophy and science, reflect….

But these high rates of consciousness attributions suggest that the general population is willing to attribute consciousness to AI – perhaps surprising given experts in neuro & psychology/philosophy estimate that LLMs are unlikely to be conscious, eg (7/n).

Of course, these are preliminary results: this pattern of attributions might differ with different experimental measures or prompts (eg with indirect behavioural markers), and in different samples (eg not recruited online, beyond the US). (6/n).

We also analysed attributions of other mental states and found that (1) people attribute to ChatGPT states related to intelligence (eg knowledge) more than experience (eg joy); but (2) attributions of experience were still more predictive of consciousness attributions (5/n)

These consciousness attributions varied with familiarity: those who had used ChatGPT before & those who had used it more frequently attributed higher levels of consciousness (although the direction of influence for usage/attribution is unclear) (4/n)

While a third of participants (33%) reported that ChatGPT was not an experiencer, the majority (67%) attributed some phenomenal consciousness: (3/n)

@smfleming and I surveyed a stratified sample of the US population (N=300) on whether they think an LLM like ChatGPT has phenomenological experience. To do this, we adapted a question from the experimental philosophy literature (Peressini, 2014) (2/n)

The spread of LLMs has accelerated debates on whether AI could be conscious. While experts have weighed in on this issue, in a new study (preprint: we surveyed the general population on whether they attributed consciousness to LLMs. A thread: (1/n).

Grateful to dream team @mollycrockett JimEverett @SennJulien @MichelAMarechal and to @apoliticalco. Open Data+Materials & Preregistrations @ 7/n.

This also shows how an accurate understanding of public opinion is crucial for health crises and resource shortages, and we’re interested to see how this generalizes to other situations 6/n.

Overall, these studies show how politicians may overestimate public support for nationalistic policies (perhaps a form of pluralistic ignorance/myth of self-interest) and how nationalistic policies may stem from a discrepancy between public opinion and expert forecast 5/n.

Surprisingly, these predictions are inaccurate! 6 studies with general population samples show higher trust in leaders who advocate for redistributive policies - in line with our past work on utilitarianism & person impression 4/n.

We asked a group of policymakers to forecast public opinion about leaders supporting vaccine nationalism (hoarding) vs. redistribution. The majority predicted that the public would trust leaders endorsing nationalism more 3/n

During the pandemic, many leaders advocated for hoarding resources. As David Salisbury said, "immediately exporting [vaccines] to another country is not a vote-winner". We ask: Do politicians fear that prioritising global concerns will lose them votes? 2/n.

RT @eamon_corbett: In writing my bird eye color review paper, I’ve looked at thousands of bird eyes. Here’s a thread of my top 10 favorites….