The current paradigm of artificial intelligence is built on silicon-based architectures—systems that rely on standardized microelectronics and von Neumann computation models. While these systems have enabled rapid progress in data processing and pattern recognition, they are increasingly constrained by physical limits, energy demands, and a lack of adaptability to the organic and distributed nature of biological cognition.
At the Institute for the Mind and Technology, we investigate the concept of ex-silico computing: a category of emerging computational approaches that explore non-conventional substrates for intelligent behavior. This includes but is not limited to organic, chemical, optical, and neuromorphic systems—platforms where computation arises not from discrete logic gates, but from the intrinsic physical properties of the medium itself.
The interest in ex-silico systems is not purely technical. It reflects a broader shift in how intelligence is conceptualized. Rather than treating cognition as a purely symbolic or algorithmic function, ex-silico approaches consider intelligence to be embodied, distributed, and context-sensitive—emerging from the interaction between structure, environment, and dynamic feedback.
In this view, computation is not merely the manipulation of abstract information. It is a material process, shaped by the constraints and affordances of the system in which it occurs. Organic and hybrid substrates may allow for forms of problem-solving, adaptation, and learning that are inaccessible—or prohibitively inefficient—on traditional hardware.
Our work in this area is exploratory but principled. We aim to understand how alternative substrates might give rise to forms of machine intelligence that are:
- More energetically efficient,
- More structurally adaptive,
- And potentially more compatible with the characteristics of biological neural systems.
Ex-silico computing raises foundational questions about the boundary between the artificial and the organic. Can a system built from living cells utilized as a computing device?
What forms of agency or unpredictability arise when intelligence is embedded in matter that evolves, mutates, or self-organizes?
What responsibilities do we carry when the materials we engineer become, in some sense, capable of autonomous behavior?
These questions are not speculative for their own sake. They lie at the heart of what it means to build technology that participates in the fabric of life. As we explore new frontiers in brain-machine integration, the limitations of conventional computing become more apparent. A future in which machines think alongside us may require substrates that do not simply imitate neural systems, but resonate with them—structurally, dynamically, and materially.
Ex-silico computing is not a replacement for silicon-based architectures, but a complementary exploration. It represents our commitment to approaching intelligence not as a fixed model, but as an open field of inquiry—one that remains grounded in science while remaining alert to the unexpected possibilities emerging at the boundary between matter, information, and life.