Page 119 - Artificial Intelligence in the Age of Neural Networks and Brain Computing

4. Wearable Personal Assistants 107

Media Guide. The personal assistant could also suggest to his user news arti-
cles, videos, movies, or music based on user history and current preferences
which he could then stream to any of the user’s device. Unlike the existing
media assistants which push the services of a proprietary manufacturer or
content provider, it would always act on the user’s behalf and keep his confi-
dential information protected from prying eyes.
Lifestyle and Transactions. Finally, neural assistants could suggest to the user
suitable activity and even perform many types of transactions online, from
renting a room on Airbnb, and hiring a plumber on TaskRabbit, to investing in
electronic currencies such as bitcoin. In fact, blockchain technology and
Ethereum smart contracts are perfect match for wearable assistants in their
never ceasing effort to protect the user data. Over time they become their users’
alter egos.
I conclude this chapter with some principles of the design for neural-based wear-
able personal assistants I developed over the years. Taken together, they offer a
glimpse of a new way of not only looking at AI and personal assistants, but perhaps
they also offer a new view of how our brains and bodies think and act in our world:
1. Neural Populations Form Fields, Not Structures. Structures, like neural
networks, are static spatial patterns, frozen in time. Resonant neural populations
form and reform dynamically, in real time, over the entire brain via AM carrier
waves.
2. Meaning Is Embodied. Meanings reside in the mutually entrained ecosystem of
brain, body, and environment. They are not representations; meanings arise
from the global resonance of large neuron populations in the cortex. They are
assembled from already meaningful local multimodal sensor streams which are
combined to form context dependent attractor basins. Meanings do not reside in
the outside world, but symbols, and other artifacts such as a knife, traffic sign,
gestures, alphabet, grammar, do.
3. Meaning Is Context-Based. When learning to walk, infants will produce walking
behavior earlier when placed in inductive context, that is, buoyed by water, or
put on the treadmill. Meaning of the word “bank” in “I sat on a river bank,”
differs from its meaning in “I walked into Citi bank.” Same color swatch will
look different on different colored backgrounds, and the same music phrase will
sound different depending on what notes preceded or followed it.
4. The Brain Does Not Represent, It Assimilates. It does not “store processed input
signals,” or stimuli, there are no representations inside the brain, that is, no links
between the internal world of symbols and external world of objects. The body
does not “absorb” stimuli, but changes its own form to become similar to
aspects of stimuli that are relevant to the intent that emerged within the brain.
5. Categorization Is Perceptual. Concepts form when two or more percepts
temporally correlate. Meaning of a concept is not based on its link to the outside
world, but on its usefulness to the user, planning actions in different