Visit this link to see a video version of this abstract. At this moment in technological history, it seems that AI-powered technology has the potential to evolve into almost anything within the next 20 years. While we expect machines to don various forms of intelligence, we also expect to integrate them into our daily lives in ways we haven’t yet imagined. How will their presence and capabilities affect our everyday human experience? While we’re often (rightfully) thinking about how our day-to-day lives will change, we rarely pause to consider the experience of the machines themselves. But there’s a good reason for this. What a machine “experiences” is difficult to define, much less measure. We also have difficulty understanding the concept of experience in general. We don’t fully understand the experiences of the many other living creatures who’ve shared our world for millennia. So while we cannot yet measure how models like ChatGPT[l] or Stable Diffusion[2] experience a written conversation, we may be able to experiment with different ways of translating a machine “experience” to a human one. How do current algorithms translate their inputs into an output, and what happens along the way? In this art installation, we introduce wearable technology meant to translate aspects of what a trained model allocates attention to into something a human can experience.

In this paper, we wish to investigate the dynamics of information transfer in evolutionary dynamics. We use information theoretic tools to track how much information an evolving population has obtained and managed to retain about different environments that it is exposed to. By understanding the dynamics of information gain and loss in a static environment, we predict how that same evolutionary system would behave when the environment is fluctuating. Specifically, we anticipate a cross-over between the regime in which fluctuations improve the ability of the evolutionary system to capture environmental information and the regime in which the fluctuations inhibit it, governed by a cross-over in the timescales of information gain and decay.

We investigate the use of attentional neural network layers in order to learn a ‘behavior characterization’ which can be used to drive novelty search and curiosity-based policies. The space is structured towards answering a particular distribution of questions, which are used in a supervised way to train the attentional neural network. We find that in a 2d exploration task, the structure of the space successfully encodes local sensory-motor contingencies such that even a greedy local ‘do the most novel action’ policy with no reinforcement learning or evolution can explore the space quickly. We also apply this to a high/low number guessing game task, and find that guessing according to the learned attention profile performs active inference and can discover the correct number more quickly than an exact but passive approach.