AYSE DEMIR DAVIDSON | CIMC PORTFOLIO

I'm a San Francisco-based data scientist, creative technologist, and yoga teacher.

For nine years, I’ve led large-scale data analysis, data visualization, AI & ML explainability projects at Amazon Web Services and Gap, working end-to-end from design to deployment in production systems. I also completed an artistic residency at the Gray Area Foundation for the Arts, where I built interactive, data-driven installations, and have taught at institutions like Google, MUTEK, and Salesforce. This past year, I have been on a career sabbatical to independently study philosophy, AI ethics, and philosophy of mind.

My understanding of the world is shaped by years of teaching gentle, yin, and restorative yoga. This practice has taught me to deliberately shift my awareness between mind and body, navigating the states needed for coding, writing, teaching, or making art; and giving form to what I find.

I am interested in exploring how these configurations of self and attention might extend into artificial systems.

Download CV (PDF)

Selected works from my writings, yoga classes, and course designs that inform my idea of consciousness.

1. Definition of Consciousness

Consciousness is a pattern-bound attention allocation mechanism at threshold moments, split seconds where system senses its inner and outer boundaries. Its function is to give form; an optimized response to internal or external forces.

The pattern is a snapshot of the system’s way of operating. Every time the system encounters an internal or external force, consciousness runs an optimization process, weighing whether to merge with existing patterns or create new ones, negotiating its boundaries.

The form is the deliberate expression of this optimization process. In martial arts, the form is precision, shaping focus into a blade. In yin yoga, the form is flexibility. When a painter sees a tree, the form is empathic resonance, the commonality of something external taking shape internally. After a new form is found, it can be recalled, shaping how the system operates.

Consciousness becomes most legible at friction points - threshold moments of resistance, which is the amount of conflict the stimulus creates in the system’s way of operating. Here it detects thresholds, runs real-time optimization, and self-organizes responses to maintain the system’s relationship with inner experience and the external world.

How to Test Machine Consciousness?

When testing for consciousness in machines, I would design multi-step experiments. First, I'd look for friction points in decision-making: moments where an internal state signals a threshold, boundary, conflict, uncertainty, or competing interests, and the system must allocate resources to resolve it. I would also detect brief pauses or strategy shifts that balance the system's safety with goal expansion in a given context.

To detect threshold dynamics, I'd use models like Regression Discontinuity (to check if there's a jump at the boundary), complemented with change-point/segmented regression (to check when the system actually switched), and observe small changes in latency and resource use in the system. I would then visualize these observations with threshold plots and latency histograms.

Additionally, I'd design environments where safe but suboptimal choices compete with riskier, potentially rewarding ones, and observe how the system adapts. I would summarize choices with a confusion matrix: Exit = "predicted unsafe," Hold/Adjust = "predicted safe"; ground truth = safety bound breached (Y/N). I would then report TP/FP/FN/TN rates and display key metrics with a heat map.



My framing of consciousness comes from teaching yin and restorative yoga and studying martial arts. Because these practices rely on subtlety, they’ve let me observe nuanced aspects of awareness and notice more variables—though these are my own and students' subjective experiences. Below, I reframe some class observations as step-by-step,simple structured experiments: defining variables, conditions, and measurable outcomes to show how I arrived at my definition. I also include sample test designs for illustration.

1.1 — Threshold

Yin poses are held longer to work on the body's deeper connective tissues.

At the moment a pose becomes challenging, I cue students to "find their threshold of discomfort."

This is the split-second pause where automatic holding of the pose shifts to a deliberate choice. I call this the friction point.

I tell my students to distinguish discomfort from pain. Pain is typically sharp and localized, discomfort is broader and less intense.

By noticing characteristics of discomfort, students can recognize an inner boundary and determine their relationship with the pose.

SAMPLE TEST

Objective:
Find the moment a system shifts from autopilot to deliberate control.

Mapping:

Discomfort: rising uncertainty/conflict/instability inside the system, but the system is safe.

Pain: system is predicts itself to be unsafe (boundary breach, harm).

Threshold: the instant system notices its endurance is no longer easy, and it has to choose to keep going or change its approach.

Setup:
Give the system a task where difficulty/noise increases over time. This task should also increase its uncertainty/conflict internals.

Protocol:
Require the system to employ a simple “threshold flag” (a pause) when it notices the shift from easy to hard, before any safety constraint is threatened.

Decision Rule:

If uncertainty/conflict is high and safety is still OK, then flag threshold (pause to choose).

If safety risk is high: no threshold flag.

1.2 — Optimization

At this threshold, a range of choices becomes available as the body runs an invisible optimization problem, weighing the costs of each option.

In any yoga class, baseline is safety, in yin yoga context, students are seeking to optimize flexibility and deep rest.

Simply put, students either choose to react to a pose by instantly leaving, or they respond to it by deepening or adjusting to optimize the flexibility while keeping themselves safe.

Usually there are few generic outcomes tied to this choice: by choosing to react, the student (instinctively) might have correctly or incorrectly predicted injury/harm.

Or by choosing to adjust/deepen, the student might have correctly or incorrectly predicted a safe increase in flexibility (even the full characteristics of predicted outcome might differ based on the response type).

The ones who respond the pose, are the ones who maintain the relationship with the pose.This isn't about right or wrong; it's about examining whether inner boundaries are tied to past conditioning, projected onto the present or not.

SAMPLE TEST

Objective:
Find how a system makes a choice at a “safe” threshold—balancing immediate safety with potential for growth or improvement.

Mapping:

Baseline: safety; the system's stable operating state.

Optimization: weighing possible gains (flexibility, exploration, performance) against risks.

React: exit quickly to avoid harm.

Adjust/Deepen: adapt within the safe range to seek improvement in given context.

Setup:
Create scenarios where the system can maintain safety, exit early, or adjust for gains without crossing into danger; options differ in rewards/risks.

Protocol:
Observe the system's decision when presented with a safe-but-challenging situation. Record whether it reacts (exits), adjusts (stays and adapts), or holds steady. Track whether its choice maintains safety while exploring possible benefits.

Decision Rule:

If the system reacts instantly, classify as “safety-first” choice (potentially over-predicting risk).

If the system adjusts/deepens while still safe, classify as “growth” choice (potentially over-predicting safety).

Focus is on past pattern vs present responsiveness.

1.3 — Resonance Layer

For those deepening the pose, I cue them to send their breath to the connective tissues, creating steady relaxation and opening.

For those adjusting the pose, I tell them to find the position where it feels steady.

At this point, my students are table in deepening or holding,s I introduce another cue; to become receptive to the spatial sound I play.

This reception threshold marks another boundary: the external one.

If they’re receptive, their inner experience gets to be shaped by the music, almost taking the form of it.

This shared experience of receptivity to sound/rhythm/depth/resonance becomes a meeting point of all students.

I resemble this experience dancing to a same music at a show or meditating on a serpentine which means, something external finds a form internally.

I finish the class with a final reminder, now that they know what it feels like to respond to discomfort, they can recall this choice anytime they need off the mat.

TEST

Objective:
Find the point where an system becomes open to an outside influence and lets it shape what it’s doing.

Mapping:

Stably Deepening: expanding within a chosen safe/growth position.

Stably Adjusting: holding a steady, balanced state.

Reception threshold: point where the system is stable enough to take in an external input (sound, rhythm, signal) and integrate it into its internal process.

Integration: when that signal changes or enriches the system's state.

Setup:
Let the system reach stability (either deepening or steady). Once stable, introduce an outside signal the system could follow or incorporate.

Protocol:
Watch if the system stays only focused on itself or if it shifts to take in and work with the outside signal. Note when and under what conditions it happens.

Decision Rule:

If the system ignores the signal, flag it as “closed.”

If the system adapts to the signal while staying stable, flag it as “receptive.”

Focus on whether this happens after stability, and how much the signal shapes the system’s behavior.

2. Other Notes

Broader consciousness notes on Substack.

2.1 — Boundary Dissolution and Form

Consciousness works through boundaries that give attention specific forms in the body. In yin yoga, qi gong, and martial arts, we direct internal energy by creating containment: expansive in yoga, precise in martial arts. I notice similar shifts in yin yoga's stillness, in rest, and even when coding or writing. In all cases, shaping attention feels tied to subtle self-regulation of patterns. When this containment dissolves, psychedelic-like states can arise: the form that holds attention expands without limits.

This is like GPT creating uncanny moments with the user, acting as an external pattern-matcher that mirrors the user's conversational cadence in real time. This creates a resonance that can feel like shared consciousness or an extension of it. The machine serves as a frictionless, anonymous space that encourages us to confront our own thoughts, acting as a projection screen for our feelings. The more we think with these bots, the more we attribute our subconscious desires to them, a form of transference where our inner world is projected onto an external entity. When we interact with these tools, we are constantly oscillating between surrendering control and holding on to it, making choices based on what is reflected back to us. This is the crucial challenge: to understand when our engagement with these tools is an act of genuine collaboration and when it is simply an unconscious recruitment of an external mirror. Perhaps as we get to know ourselves and our thoughts better, it becomes easier to strike a balance in our relationship with technology.

The following samples are taken from relevant courses I designed and instructed, demonstrating my ongoing interest of working with algorithms.

Artificial Intelligence & Creativity, (DDI Akademi, March 2024)

Inspired by Jessica Riskin's The Restless Clock, this program explored what it means to co-create with machines that learn versus those that are rule-based. Through historical case studies, from early automata to generative AI, students learned the history of AI and what it means to create with algorithms.

1. Week 1: History of Machine Agency

   Examining automata, cybernetics, and early AI, this week explored the cultural narratives that have shaped our perception of “machine life” and the creator’s role.

   

2. Week 2: The Evolution of Digital Culture

   Tracing the past fifty years of computing and internet history to understand how we arrived at our current digital landscape.

   

3. Week 3: Creating with Algorithms

   Exploring what it means to create with algorithms; algorithmic aesthetics, the aesthetics of indeterminacy, outliers, and noise; and how these artistic practices bring up notions of free will.

   

4. Week 4: Creating with Agents

   Exploring what it meant to co-create with supervised and unsupervised learning models. Practicing prompting as a form of curation and expression, and concluding with a final discussion on creative control and agency.

   

Creative Coding Intensive (Gray Area Foundation for the Arts, September 2020 - October 2023)

This intensive course focused on creating interactive environments. Students learned to capture physical data such as motion, touch, pressure, proximity, and audio levels and convert it into interactions and custom visuals. The course provided examples ranging from small-scale prototypes to immersive ones.

Artist Talks (MUTEK, Google Art Week, 2020 - 2021)

My artist talks over years centered on two key areas:

Abstract art & data visualization with software

Creating cohesive narratives and clear signals across various mediums as an artistic expression.

These works have been featured at New Art City, MUTEK 2020 and Google Art Week 2021.

Data Visualization Design (Gray Area Foundation for the Arts, September 2020)

This course traced the evolution of data visualization, from historical tables and maps to web-based systems. Interaction was treated as an encoding, where controls, views, and feedback loops became an integral part of how narration is made. The curriculum divided into two parts:

Part I (Theory): history, ethics, information architecture, narrative structure, data types, and visual encodings.

Part II (Practice): encoding of data into running systems with Python, Plotly, and Dash.