AI, Anima, and the Uncanny

Unease about artificial intelligence is often expressed in terms of the effects of bias, or job displacement, or singularity (the fear that AI will take over). Our thesis is that a major source of uneasiness, one which has not been much explored, is AI’s uncanniness.

AI, Anima, and the Uncanny

This website is our submission for the Religion and AI Seminar at the AAR/SBL 2019 Annual Meeting in San Diego. We consider this the beginning of a platform for our research into why artificial intelligence is a major source of uneasiness for many people. In short, it is our belief that the notion of “uncanniness” provides explanatory potential for analyzing how humans experience AI. Furthermore, we suggest that Aristotle’s concept of anima affords insight into the uncanniness experienced by many humans when they interact with AI.

In addition to the homepage, this site comprises five elements, which can be accessed by clicking on the corresponding menu items listed above or the links below. They are briefly summarized here for the purpose of orienting visitors to this website.

1. Thesis

Unease about artificial intelligence is often expressed in terms of the effects of bias, or job displacement, or the expectation of singularity (the fear that AI will displace humans as the dominant species on earth). Fears about AI almost always focus on AI’s superior intelligence, because such a super intelligence will be beyond human control, and AI, in whatever form, will inevitably “outsmart” us and pursue its own ends. Our thesis is that an unnamed source of uneasiness, which is masked by the terminology of “intelligence,” is that humans experience AI as uncanny. This pre-cognitive uncanniness, however, arises because AI is fundamentally familiar and unfamiliar at the same time; like us and unlike, us it challenges our understanding of human nature and what it means to live a human life in relation to non-humans.

2. Topic Model

In addition to the familiar academic modes of research and interpretation of text, we deployed unsupervised machine-learning as an experiment in using digital resources that are commonly regarded as simplified or proto forms of AI technology. More specifically, we utilized topic-modeling to analyze selected twentieth century philosophical articles to test whether mind-body dualism is as stable and persistent an idea as we hypothesize. The model we used is rudimentary, so it serves more as an illustrative instrument than a conclusive one. On this page we provide a visualization of our results as well as a link to the github page where our project resides and where you can find the code we wrote to create the topic model.

3. GPT 2

GPT2, the much discussed brainchild of OpenAI, has been described has a form of language processing capable of finishing your sentences for you. Here you will find a brief description of GPT 2 along with a link to a site where you can try it out for yourself. We include this page as part of our presentation because it enables visitors to experience for themselves a machine capable of authoring its own text with only miniscule input from a user. We hope its inclusion will inspire discussion about whether AI thinks for itself, if that thinking is similar to or different from human thinking, and what kinds of reactions that can occasion.

4. Conclusion

By way of conclusion we offer very brief comments about where we think this line of inquiry will lead and questions we might pursue when we come together at AAR.

5. Bibliography

Here you will find a working bibliography, which has served as a launching point for our thinking on these topics. The literature on AI is growing exponentially so we do not consider this bibliography to be more than a starting point.

Pamela Eisenbaum

Micah D. Saxton

Theodore Vial