Artificial Intelligence

when a.i and neuroscience collide.

 If there were machines which bore a resemblance to our bodies and imitated
our actions as closely as possible for all practical purposes, we should still have
two very certain means of recognizing that they were not real men. The first is
that they could never use words, or put together signs, as we do in order to
declare our thoughts to others. For we can certainly conceive of a machine so
constructed that it utters words, and even utters words that correspond to bodily
actions causing a change in its organs. But it is not conceivable that such a
machine should produce different arrangements of words so as to give an
appropriately meaningful answer to whatever is said in its presence, as the
dullest of men can do. Secondly, even though some machines might do some
things as well as we do them, or perhaps even better, they would inevitably fail
in others, which would reveal that they are acting not from understanding, but
only from the disposition of their organs. For whereas reason is a universal
instrument, which can be used in all kinds of situations, these organs need some
particular action; hence it is for all practical purposes impossible for a machine
to have enough different organs to make it act in all the contingencies of life in
the way in which our reason makes us act.
Descartes – Discourse on the Method (1637).
overview.
four essential questions:

intelligence: what is it?

●

'true' AI?
●

how does neuroscience play a role?

●

implementation?
●
#define intelligence;

What does it mean for a non-sentient entity to be

'intelligent'?

DNE.

Alan Turing and the foundation of modern

computing.
#define intelligence;
Turing test: if an intelligent machine can convince
any human observer that it is intelligent, then the
machine is intelligent.
(the imitation test.)

Simplistic?
 #define intelligence;
knowledge representation.
-Qualification problem.
-Sub-symbolic representation.

planning.
-decision-making.

learning.
-relates to knowledge representation, but
how?
 modeling the brain.

Symbolic or physical.

Research into neuron-clumps wired to perform

basic motor tasks (zombies?)

Most efficient: electrical basis.

What level?
-whole brain?
modeling the brain.

Blue Brain Project.

-models individual neurons and synapse firing.
-biologically realistic.

Blue Gene supercomputer.

-8192 processing cores.

Completed model of rat neocortical module in

2006, 10^4 neurons and 10^8 synapses.
modeling the brain.
Neural networks
(symbolic)
● Computational units
capable of learning.
● Modifies output based on
feedback (comparable to
LTD and LTP.)
● Implementation of fuzzy
logic.
● Capable of rerouting
itself to find the 'optimal
path.'
modeling the brain.
Neural networks, continued.

Application of fuzzy logic (FNN).

Computationally inexpensive (compared to a

BG/L.)

Requires learning (training) – learn by example.

Supervised or unsupervised.
modeling the brain.
Another approach:

Example: immediate recognition (vision)

Passing data through alternate filtering and

integration layers (similar to the visual cortex).
-building cognition from the ground up.

-self-training, emergence of basic Boolean logic.

Increasing accuracy, performs better than human
test subjects (!)
apply intelligence;

The Open Mind Common Sense project tries to

solve the common-knowledge problem.
-Is a combination of a four dimensional mapper
and a FNN (it just works. Don't ask how.)

Uses bits of common knowledge and extrapolates

(somewhat like fuzzy logic, similar to the concept
of assuming.)
apply intelligence;

Decision-making based on uncertainty has been

observed to model Bayesian probability.
- a Bayesian model is primarily a statistical model
with constant updates from neural processing.

Modern intelligence processing uses a Bayesian

model on some level.
stumbling blocks.
The fundamental difference between brains and
computers.
-Massive parallelism vs linear computation.

Processing is not differentiated from memory.

Biological variations (different cell types, even

slight mutations) contribute to a massively
complex system.

(we're getting there, though most likely not by

2030.)
 Bibliography.

http://www.technologyreview.com/read_article.asp
x?
id=17111&pg=5http://www.neuro.gatech.edu/grou
ps/potter/papers/Potter-
NeuroscienceForAIchapter.pdf