Rene Girard & Mimesis in Artificial Intelligence

Girardian Thought, Important Model for Encountering Artifical Intelligence

I have said before that AI is nothing other than thoughtware masquerading as software.

What do I mean by this?

There are some who will look to AI and see only the machines which are learning. For these AI is a matter of code and programming. It's about compute operations. The content is important not essential. That the content is correct is important but not imperative. Their QA is for function, qualitative to an extent but binarilly quantitative -- not content, a matter of the technical operations enabling the subject of their work.

For others, myself included, the code aspect is therefore completely secondary. We do not care, and are unimpressed by the technical operations ("QA it, fine, whatever. Does it work? Can I SEE it?") and more concerned with AI as a new marvelous digital expression of our content and ideas. For us the content is everything, our interest is in accessibility, not operations, so long as it operates.

So how we look at things matters. Ontology matters. The things we know and understand as what things matter. What are things? And this includes technology, and this includes AI.

Also our framing of our understanding matters. Not just what we believe things are, but how we come to believe and experience what they are. The "lens" through which we see can be chosen by us or (as maybe in most cases) chosen for us.

Here's a proposal for a frame to better understand artificial intelligence; what it is and is not, how we encounter it, and what it represents to humankind.

One of the minds I came to knowing about late (in fact only after I had the providence to spend an hour sitting with he as an unknown and a mutual friend talking together -- only obtaining his books and accessing his thought later, after the fact), is a peerless thinker for our times - Rene Girard.

René Girard (1923-2015) was a French historian, literary critic, and ultimately a philosopher.

Born on Christmas Day in 1923 in Avignon, France, Girard grew up during the German occupation. After completing his studies at L'école des Chartes in Paris, he moved to the United States in 1947 to pursue his academic career.

Girard received his PhD from Indiana University in 1950 with a dissertation on "American Opinion on France, 1940-43". He taught at various institutions, including Bryn Mawr, Duke, and the State University of New York at Buffalo, before joining Stanford University in 1981 as the Andrew B. Hammond Professor in French Language, Literature and Civilization.

Throughout his career, Girard developed new influential theories on mimetic desire, the scapegoat mechanism, and comprehending human behavior. Despite literally groundbreaking work, he viewed himself only as a messenger of what was true already.

He wrote 30 books including some which really touched me and others deeply "Violence and the Sacred" (1972), "Things Hidden Since the Foundation of the World" (1978), and "The Scapegoat" (1982).

It would be impossible to sum his thought into one article and so I choose to highlight some of his most prominent original gifts of thought to humankind instead.

René Girard's teachings center around three concepts: mimetic desire, and two related concepts the scapegoat mechansim, and revelation.

1. Mimetic Desire - Girard proposed that human desire is not linear or autonomous, but imitative in nature

1. People don't inherently know what to desire; we look to others for cues and imitate.
2. We desire objects firstly because others desire them, not because of objects' inherent value.
3. There is a triangular relationship between subject, model and object of desire.
4. This imitative desire forms a triangular relationship between the subject, the model, and the object of desire.

Consequences of Mimetic Desire - Rivalry and Violence

As a consequence of mimetic desire:

1. Imitation can lead to conflicts when people desire the same objects.
2. Models can become rivals, competing for the same desired objects.
3. This rivalry can escalate, potentially leading to widespread conflict.

2. The Scapegoat Mechanism

To prevent total societal breakdown due to mimetic violence, Girard proposed the scapegoat mechanism:

1. Communities unite against an arbitrary victim, blaming them for the chaos.
2. The expulsion or sacrifice of this scapegoat temporarily restores social order.
3. This mechanism forms the foundation of human culture and ritual; and it repeats. Often yesterday's scapegoat ascends to be today's persecutor. And yesterday's persecutor likely eventually becomes a victim.

3. Revelation

Girard argued that the Judeo-Christian scriptures reveal the innocence of the scapegoat victim, exposing the unjust nature of collective violence, as it is a mechanism of the whole society nevertheless instead of a fault in the victim. This is new thought, because ancient and post-ancient societies embranced "what we always believed" that "he had it coming to him."

It challenges many established ideas in various fields, including psychology, economics, and anthropology, by emphasizing the interdependent nature of human motivation and behavior, the "necessity" for society of creating scapegoats as a "pressure valve" (my words) for averting larger wide-scale violence, disorder, and breakdown. And yet (and in the more) is the scapegoat innocent (in the abstract) 'performs a service' (unwillingly) "for society" -- separate from his own culpability.

What has mimetic desire, scapegoating, and revelation have to do with better understanding and encountering AI? There are many stunning parallels.

Mimetic Desire

For one thing, the content we share (inputs), and the content we seek to share (outputs) (be these in prompts into an LLM or another AI tool), represent our expressions of what we seek from the technology. In a sense we must remain aware of the nature of mimesis when encountering 'synthetic near thought.'

Because, since we are readily susceptible to easy imitation, we can be too easily influenced both in what we say, and in what we read. AI, as an "artificial mind" nevertheless simulates our thinking very very well, and so has the power to influence us in ways we would not imagine were we not to understand mimetic desire.

One good example of this is in suggested prompting. We see this already for years in Internet search tools like Google. What we are suggested to look for; word or phrase, become really and actually the things we are "thinking." And since AI has infinitely more parameters than even the Internet, content and context and words matter much more deeply.

As the subject, are we seeking objects which are nothing other than the model of some mediator we have thoughtlessly "chosen" to imitate?

It is the same with outputs. If our questions (prompts) are structured for us ambiently, these questions become a component of interpretation by us of the answers provided.

And so if I am lead to ask for "fifty fish" and given a reply of a bag of candy Swedish Fish, I will not only be thinking of Swedish Fish, but will also be thinking that Swedish Fish may be more essential than the question/thought I had in the first place.

The natural conflict wrought by mimetic desire will be entirely electronic and in our minds. And 'alone and in secret' we tend to be most powerfully influenced. And since we do not remain alone and in secret, these formations are brought by us into the wider community participation.

What I am saying is that it is hard not to see this way the immense power of influence that is AI.

The Scapegoat

In AI the notion that what is popularly-held is correct is actually a human notion which will be defended irrespective if one is in the technology context or not. A "democracy of knowledge" will of necessity accord with "outsider theory."

We have seen AI-adjacent (electronic human communications) events like the example of rapper, designer, and provocateur Kanye West, adding imflammatory content on X of all places. While it is clear to many that he is using infamy for attention, the human impact (on he and the community - society) remains.

Be they ugly abhorrent or stupid (or highly insightful or creative), these are only digital characters on a screen. And on X, a platform which says users can share whatever thoughts (so long as not illegal) in perfect freedom. Nevetheless, many have called for Kanye West to be "banned" or even much much worse.

It's only a little leap from the auto-scapegoating of Kanye West, to the AI environment.

Will people use AI to create salacious images of celebrities or even neighbors? Will false content about a person be fed into an llm's database. Will content which is true in an LLM be "restated" in order to disadvantage a scapegoat? Is it already happening with a President, or Elon Musk, or Palestinians, or "The Jews"?

Each/all of these have happened already. And so being aware of the "necessary" scapegoating of others in society can help AI users and designers but making each and all of us more circumspect. Since AI will touch on all of "humankind's datasets," past, present, and future, it makes more than a little sense for more of us to think in a rare and Girardian way about the human mechanism of scapegoating. And it repeats. And yesterday's scapegoat can be today's persecutor.

Finally, Revelation

The Revelation is the inherent violence undergirding society, and the "extra" (often arbitrary) danger and this great innocence of the victim/scapegoat.

Thinking in this way, about AI, presumes that we may be naturally more inclined (rather than less) to engage in scapegoat assignation.

And this is not only personal to individuals, as above, but in AI has a much wider scope. Using Kanye West as an example (scapegoat - ha!) again, we may be inclined to decide that there are "categories of information" or "sources of information" or "topics of information" that we must destroy for "the common good."

• Anything from Kanye West (e.g.)
• Anything from [Religion X] because will be "biased"
• Anything which "is insensitive"
• Anything which is "biased" (how could we either collectively or individually know?)
• Anything from "despots"
• Anything from "discredited regimes"
• Anything which is a "hate crime"

Excluding content on these bases unknowingly/automatically impoverishes AI and not only its users but only reveals the violence done to the content which belongs to mankind, but finding and eliminating "data scapegoats."

The thought of Rene Girard has very deep import for our current constructions of universal global artificial intelligence. And we should take heed.

One last great example; again not AI, but immediately translatable because it represents human behavior and "data choices" is the Gulf of Mexico / Gulf of America spat between the White House and the Associated Press.

The name, really and actually and legally and officially, has been changed to Gulf of America.

And yet, because they either do not like it, or more precisely do not like the man who did it, or even more precisely do not like that he has the power to assign it legally, a press agency (which is supposed to be, ontologically, a conveyer of what is) chooses to continue to say what is not, rendering themselves a scapegoat in attempt to render another.

One can thus see how fast (instant) is factual chaos to break out in society, and why our AI should be maximally safeguarded through smart awareness.

#renegirard #scapegoat #mimesis #mimetic #kanyewest #artificalintelligence #ai