Embodied Personal Assistants
Mobile personal assistants (PA) should guard the customer data with their life; they should not be designed to share customer data with advertisers as they do today. Also, the vast majority of today’s PA’s such as Siri, use AI-based IP metaphor to approximate the meaning of the user actions. After 20+ years of brain and PA research, four of those with Walter Freeman, the seminal neural scientist at Berkeley, I have come to believe this metaphor does not work.
Metaphors matter; once one realizes that brains did not evolve to process information but to support intentional human behavior in novel environments, the static neural networks become less important. Instead, dynamic e-fields created by resonant populations of neurons embedded and embodied within the moving body come into focus.
Without understanding the user’s context and his intentional motor behavior, “personal assistant” with all its “symbolic models” will inevitably fail. Once, when Napoleon Bonaparte was asked for advice on how to win a battle, he answered: “You jump in and see.” He was right; the user’s intentional action within his present context is the essence of embodied thinking. Disembodied thinking, such as Tony Fauci’s Covid-19 symbolic models, can not produce real understanding.
My company, Embody Corp believes three critical ingredients are missing from the existing personal assistants, preventing them from understanding the user:
- AI neural networks comprised of synapses linked by slow axons are augmented in the human brain by tiny neuron oscillators forming large, resonant, and lighting fast e-fields in theta, beta, and gamma range, supporting all meaningful actions. We recently discovered an emergent hardware technology that can deliver these fields inexpensively on top of existing CMOS neurons.
- To truly grasp the meaning, a personal assistant must be grounded in the embodied motor action in addition to symbols. It turns out that all wearable sensors are motor sensors, including vision and hearing sensors. AI engineers do not understand this, as can be seen from current AR/VR static displays, which perform poorly because they ignore the motor aspect of human perception, such as eye saccades. 50% of proprioceptors in the human body are in the neck and ocular muscles.
- Any personal assistant must protect the privacy of valuable user memories. The best way to do that is with the internet of values, i.e., blockchain.
The recent discoveries of field phenomena, such as hippocampal phase precession and sharp-wave ripples present during long term memory formation projected from the hippocampus to the frontal lobe, already play an important role in Google’s Deep Mind work. Embody Corp assistants take this approach to a whole new level. When embodied thinking is finally coupled with the emerging e-fields hardware, personal assistants capable of meaningful user understanding will become possible. They will become our trusted advisers; indeed they will become our alter egos.