It's Not Artificial Intelligence

It is now the everyday vernacular of my existence and that of many professionals in financial services. “Artificial intelligence and machine learning”- the new enlightened water cooler and cocktail party enduring topic of conversation.

Why is it called artificial? What is intelligence? How can a machine learn without a brain? 

Origins of the Term 

Tim Skene, Mechanical Engineering & Product Designer of Physical Goods, Carleton University, states that the term was first coined in 1956, when a gathering of computer scientists conducting a workshop at Dartmouth College, published the phrase to  describe their intent to  learn how to give computers human-level capabilities. The originator of the event referenced “artificial intelligence” as the mission of the collaborating scientists.

How is it Artificial?

Why artificial? Why intelligent? How can intelligence be fake?

A  better term, not as "catchy" would be  synthetic cognitive responsiveness. 

Human beings possess actual chemical cognition, the miracle of the bio-chemistry and evolution of the synapses of our brain function that enable us to conduct mental processes to gain knowledge, comprehension, meaning---resulting in remembering, judging, problem-solving, imagination, perception, planning and execution. 

Machines on the other hand, cannot imagine, dream or aspire. But they can learn and judge. They can be electronically engineered to measure  patterns in data and apply mathematical algorithms to deterministically recommend an action, initiative---whether that's producing a value, exercising an order of operations, or triggering permission for a digital process to start, machines can actually simulate human determinism. 

Machines cannot behave illogically. They cannot expand their imagination, or dream, but they are excellent for  computing probability, processing equations, comparing reference data, and presenting  alternatives as prioritized true/false statements. 

And that is a very useful friend to have.

Failure  is the Foundation of Intelligence

Putting aside the banter around AI, what is most fascinating to me is the learning process that builds cognitive agility, intelligence and better judgement and action-taking.  In that way, learning is very similar in organic and inorganic entities. Intelligence is derived from failure. 

Knowledge is the learning genome, the reference architecture of facts, codes, symbols, instructions that we and machines are both capable of archiving.  Knowledge is insufficient for intelligence. Knowledge is just stored and codified data, it can't be intelligent until it is manipulated, mutated, coagulated and utilized as action. 

We can't do an inventory of every single thing we know or possess in our data base when we need to take action. Trial and error is the author of intelligence: “Ouch, that trial hurt…that didn't work…it caused me grief…pain. On the flip side, this trial produced a positive experience, outcome, feeling, one I would like to replicate. 

Learning requires the mutation of applied stored knowledge variants, as a trial action, then as an actual consequence. Tacit intelligence accumulates when it becomes automatic to consistently replicate behaviors, achieving  predictable results within a negligible range of variation, reinforcing a decision to elect one alternative to no alternative from a host of competing options.

AI and the Culture of Fail Fast

Before we can embrace AI we need to acknowledge that learning is messy, sloppy, sometimes inefficient, painful, boring, inconvenient, tedious, expensive, time consuming and maybe even dangerous. 

But mechanical and biological learning systems all learn by failing as an antecedent to continuous improvement. Continuous improvement leads to exceptionalism. Intelligence is knowing how to produce exceptional results.

Here are five pre-requisites for any organization that has a mission to deploy AI Machine Learning across their enterprise for the sake of quality, productivity or distinguished service. The following must first be true, or the organization could lose its’ conviction and or commitment:

1.     Failure must be rewarded, expected and celebrated within the organization.

2.     The opportunity to experiment, try, simulate, prove ideas must be a daily norm.

3.     Knowledge, therefore data, must exist in organized, archived abundance.

4.     Data must be readily accessible to everyone at every level of the organization.

5.     Budgets have to appropriate funding for innovation and experimentation equivalent to allocations for risk management and process control.

There is nothing artificial about intelligence. It is learned responsiveness to a reinforced best action decision, creating coded instructions for reproduction and replication.  Failure is the developer of that code.