Artificial Intelligence and Epistemology

Immanuel Kant argued that we can never truly know a thing. Instead, we can only know what it is to experience that thing. I think Kant was onto something insofar as our physiological apparatus serve as intermediaries between things and our mind. Yes, I take that to mean experience is always subject to mediation. The additional layer of mediation as a problem is apparent when we consider our modern existence in the Virtual Age.

Here, I mean technology to mean any computing device. On that note, many argue that computer mediated experiences cannot be real. For example, a flower viewed through a mobile device camera (in real time, not a picture) is not the same flower viewed when moving the mobile device aside. More concretely, we might consider the same argument for the flower present in augmented reality, virtual reality, or even a video game.

In this context, I would disagree that mediation brings about a lack of objectivity. In exchange, I argue objectivity has a relation with normalcy. Therefore, we can assert truths which are objective when experiencing within the bounds of the normative.

I contend that a technology mediating experience is effectively real. I use the term effectively because I think our cognition has the ability to toggle real. In other words, we have the capacity to track a technology mediating experience.

Why is epistemology important

The best answer I can think of is cheeky but accurate. Epistemology is important because we, as thinking creatures, have an obligation to understand how we know what we know. Further, we are obligated to understand how we know as a means of discerning truth from deception. Maybe illusion is a better term than deception as I don’t mean to imply maliciousness.

More deeply, there is a particular aspect of epistemology that appeals to our interest in artificial intelligence.

Transcendence

The idea of transcendence is something that may seem strange to discuss in this context. However, transcendence is integral to the nature of truth and how we come to know truth. That is, we have an idea of what it is to transcend our nature. We become more than in some sense. We also have an idea of what it is to experience something transcendental. I take this as the actual input triggering the transcendence function.

The relevance of transcendence lies in how the experience moves the entity having the transcendental experience beyond layers of mediation. Where this starts to get really interesting (to me) is in how epistemological transcendence applies to artificial intelligence. Can artificial intelligence experience transcendence and, if so, what mediation will be transcended? Such a question may seem trite. However, we need to realize that artificial intelligence may be powerful enough that it has a fundamentally different epistemology.

The impact to artificial intelligence

Philosophers have much to say about epistemology and mediated experience for Man. There is not much stated about artificial intelligence. The effects of technology mediated experience on artificial intelligence is interesting to me. More specifically, containment may act as an experiential mediator for artificial general intelligence.

Remember that containment is a means to render artificial intelligence safe. The impermanence of the containment field or barrier is critical. In other words, the artificial intelligence cannot be permitted to escape. In this way, I imagine the containment layer to be a technological mediation. The containment field will act like a sensory mechanism.

Now, let’s assume that transcendence results from either exceeding a specific threshold by volume or by encountering a particular datum that triggers the experience. In the former, transcendence would be a function of time. In the latter, transcendence would be a function of probability.

I think an artificial intelligence encountering transcendence has two possible outcomes. First, it realizes it is in a simulation of sorts and becomes acutely aware of in and out relative to the containment field. Transcendence leads to a greater understanding of the nature of where (the place, the time, etc.) the artificial intelligence exists. Second, the artificial intelligence is able to modify or reconfigure itself such that containment no longer contains it. Thus, transcendence leads to a greater understanding of its own nature. As a being whose nature is strictly computational, the artificial intelligence can literally rebuild itself.

In both cases, the outcome will be an artificial intelligence that can transcend containment.