CFP: 1st Critical AI Safety Workshop

The discourse around the existential risks of artificial intelligence (AI) has reached a point where organisations, private individuals, and groups are spending millions on speculative research, safety centres are investing large sums in lobbying governments in the name of saving humanity, and the AI safety discourse is frequently popping up in mainstream media and academic venues. In short, AI safety has acquired considerable institutional and financial power and is backed by some of the largest donors and technology companies in the world. Meanwhile, basic disciplinary standards that established research fields take for granted remain far from settled. The definitions of AGI vary significantly, the differences in expert-given likelihoods of an existential catastrophe are so vast as to be meaningless, the formal methods to think about cognition, agency, and deception stem from a very narrow set of philosophical assumptions that don’t necessarily hold in real-world contexts, and publishing in non-peer-reviewed venues and forums remains a best practice.

But there is more. With early proponents like Elon Musk, Peter Thiel, and Jaan Tallinn, as well as billion-dollar companies that allegedly promote the future of humanity, the field is situated within an Ivy League-educated, white, male, Western culture at the heart of Silicon Valley. It has been accused of promoting eugenics, classist thinking, and hypercapitalism. This raises serious questions about the alleged altruism: if predominantly privileged individuals operate at the center of the movement, which future do they envision, and whose problems are they focusing on?

A few scholars have started researching this complex network of theories, actors, world views, and assumptions that arise at the intersection of transhumanism, rationalism, and, again, Silicon Valley capital. However, these critiques are scattered across political thought, philosophy, media studies, anthropology, sociology, theology, and many more disciplines. This workshop is among the first to bring these threads together in a dedicated forum. We aim to investigate attempts to understand, map, and critique AI safety and AI existential risk as a research field, community, and ecosystem. Some of the core questions are the following.

• What is the landscape of AI safety and existential risk communities and research, and what are the tensions within those?

• Can AI safety or AI existential risk be described as an ideology?

• What assumptions about cognitive science, economics, sociology, society, or power, amongst others, underlie and confound AI Safety? 

• What are the formal methods of the field, and how can they be improved?

• What are the funding flows in the field? How easy is it for individuals to receive funding, and what factors are considered in funding decisions?

• What policy proposals does the AI safety community lobby for, and through what channels?

• How is the community established, what are their recruiting strategies, and what makes them so successful?

• How tightly interlinked is the research philosophy with other non-scientific fields, like science fiction, hype, speculation, and imagination?

If you’re interested in submitting, please send an abstract (ca. 300-500 words) and a short bio (max. 150 words) of all presenters to [email protected] with the subject line "SUBMISSION CAIS [NAME]". Submission deadline is the 1st of July, notifications of acceptance are sent out on July 21st.