Imagine there is a padded room with a single table on which a notepad, a tome of rules, and writing instruments rest. A single person with too much time on his hands resides in the room. Suddenly, a coded message enters in the inbox. The person gets to work, translating the inbound coded message into another outbound coded message by rote, following the rules in the tome.  Said person finally places the outbound message in the outbox.

Question: does the person in the room have any idea what he just did? If the coded message were simply a Chinese email, can we say the person in the padded room understood Chinese? Is the person intelligent?

This is Searle's Chinese Room. Dr. John Searle is a philosopher who first posed this thought experiment as a means of defining and exploring what is intelligence, what it means to be intelligent, and where precisely lies intelligence. This could have been a seminal thought experiment along the lines of the Allais Paradox that spawned the study of Behavioral Economics and Prospect Theory.  Instead we get yawns and the arbitrary split of strong AI from weak AI when the idea was apparently neither about AI nor an esoteric academic theory but about an existential question about everyone's being. 

It takes little risk to say there is universal agreement that a boot is not intelligent. Neither is a foot, nor an elbow.  But a person is intelligent and the focus centers on the brain.  What then exactly is doing the intelligence in the brain and how does it work?  If a body is a ship, complete with pipes and wires and engines and fuel stores, with a head as the command bridge, then where is the captain?  Where is the homunculus? 

The brain scientists have mapped out neural pathways connecting senses to brain region, from layered cortex to numbered Brodmann Area.  Prevalent among the senses (and the research funding) is the visual pathway.  Coded light wavelengths pass through the pupils, excite the rods and cones, and stimulate the optic nerve bundle.  Shape, motion, and colors pass through the lateral geniculate nuclei, cross over to contralateral hemispheres, diverge to affectionately labeled what and where pathways, and reconverge to the visual cortex where they pass up and down the six layered cortical columns at the back of our heads to recompose the image.  Of the region that comprehends the recomposed image there is no trace.

Still no homunculi.  The captain is not on the bridge. We aimed for Searle's Chinese Room and we got Searle's Chinese Inbox. 

What say the computer scientists attempting to turn calculators into parallel intelligence systems? In defense of Artificial Intelligence, these scientists brush off this thought experiment as irrelevant and out of scope. Just have the guy memorize the rule book, and they both would be intelligent. The tome has the artificial intelligence. The tome is the algorithm.  Computer science is about the algorithm and if the output is correct and in O(log n) time, then it matters not what happens in the black box module. The person in the room is the processor and computer science is about abstracting processor, instruction set, and data modules into independent layers. 

Still no homunculi, but if the captain is not on the bridge, we can outsource him with another.  We aimed for Searle's Chinese Room and we got Searle's Room in China.

Various other attempts include robotics approaches to embody the person in the room to have arms, legs, and sensors outside of the box to receive feedback and associate the outside with the internal tome of rules, notepad, writing instruments, and coded message. While intriguing, these attempts cloud the issue by violating the constraints of the experiment. One might as well invoke supernatural gods and the magic of an undiscovered mythical homunculi. One can even call it the strong AI as opposed to the weak AI.  Since both are AI, it is (a very hand-wavy) problem solved.  Then one would say Searle is just splitting hairs.

There is still no homunculi, we have no idea where is the captain, and what was the question again? 

The question is a philosophical one about what is intelligence. The question is a practical one that every child student asks at some point - "Why am I studying Math, or Physics, or Literature, or History?  I am not learning anything anyway, just memorizing the facts to pass this test.  Then would not the test ultimately be testing the book and not me?" The question is a business one that every manager asks in hiring and retaining staff, as Rolf Breuer famously termed in the War for Talent. The question is what is this intelligence, this je-ne-sais-quoi-but-I-know-it-when-I-see-it quality. Imagine finding a partner employee who can see at once what needs to be done, does it, and adjusts as needed on the go to capture unanticipated opportunities. Then we should most definitely pair up with this perfect partner. We are not looking for the slavish automaton with too much time on his hands, nor the tome of rules, nor presumably the notepad and writing instruments.  We are looking for intelligence. So where is this homunculus?

Based on years of experience in business plus years of experience in brain, behavior, economics, sociology, computers, and technology research, here is a wild but hopefully educated guess:

Intelligence is not a noun.  It is a gerund.

A noun is static. A gerund implies action and dynamic activity. Intelligence is extradimensional. No, not in a spatial dimension  with teleporting aliens, but extradimensional across time.  In Searle's Chinese Room, the person does not demonstrate intelligence. The tome does not demonstrate intelligence. The notepad and writing instruments do not demonstrate intelligence. The padded walls do not demonstrate intelligence. The entire Chinese Room does not demonstrate intelligence.  An intelligent person sending a Chinese message to the room and getting a cogent, apparently insightful output response cannot conclude the room or any entity thereof is intelligent. Just as it takes two points to provide evidence of a line, or three to imply a plane, it takes more than one response point in time to establish intelligence. Hitting a bullseye once is not accuracy.  Hitting it repeatedly as it moves under different conditions trying to dodge is.

Within the bounds of the thought experiment, if the Chinese Room continuously adjusts its responses to fit a series of questions themselves adjusting to the Chinese Room responses - i.e. a conversation - then it would exhibit intelligence. In business terms, the room would need to show it can see at once what needs to be done, does it, and adjusts as needed on the go to capture unanticipated opportunities to be an intelligent partner. In computer science terms, neither the processor nor the algorithm nor the computing system are intelligent. The intelligence lies in the business analysis requirements gathering that identify the problem task to be solved - i.e. the question and goal setting, not the solution answer. In brain science terms, neither the brain nor the neurons nor the synapses are intelligent. The growing - another gerund - of neurons in response to the inputs is intelligence.  These answers do provide concrete clues as to where to look for what precisely is intelligence.

Namely, though it takes much risk to say so, a boot is still not intelligent, but a foot technically is intelligent since it grows and shapes itself to adjust to its needs.  If the foot fits...