linkedin facebook twitter rss

17 Mar Intro to Cybernetic Models

Cyber Seeds

Are you my Motherboard?

From Eliza to WatsonJeevesSiri and Alfred, people have been naming their computers. Some systems are named after their inventors – like Wolfram Alpha and its siblings. Who is JARVIS? Just Another Rather Very Intelligent System? These attempts have, in their own little ways, touched our lives. But we are still waiting on the cusp of an innovation that will mark the end of the information age and the beginning of the age of knowledge. In this section I will be addressing specific automated approaches to mimicking human communication and other cognitive processes. I will do so in the context of automation that could be useful in modern business settings.  “Some business problems require new thinking and technology to provide the best solution” (McCreary 2014 p.8). This “new thinking” is what interests me.

Besides pontificating on my perceptions of what’s good and useful in automation and AI, I will provide many links to other web content, so you can look and decide for yourselves.

Understanding Context Cross-Reference
Click on these Links to other posts and glossary/bibliography references

  

Section 7 Intro

AI Modeling Section Icon

 

Table of Context

Computers that Understand

AI Guy 3Both my wife, Ann, and my sister, Diana, work with kids with special needs, many of whom are not super proficient at communication tasks. Diana once described how challenging it is to teach them to paraphrase text. Arguably, humans are engineered from the top down to be able to communicate, and yet it comes harder for some. How much more so for computers. High quality text summarization has long been a key goal for Natural Language Processing (NLP) systems.

There are emerging technologies that will certainly make our task easier. Expert systemsontologiesbig data, advanced speech recognition, service-oriented architectures, and other important innovations all give us good clues that can lead to smarter systems. The age of knowledge will be characterized by people relying on their electronic devices for actionable knowledge, not just useful information that they are forced to analyze before making a decision.

The Intelligent Calculator

The pocket calculator is a simple example of a machine that mimics one sort of cognitive function. Many people have decided that calculators do not qualify as artificial intelligence because even a wooden abacus can be used to do math. Although math has been simulated on machines, it still requires intelligence, as all of us who have taken advanced math classes can attest. To some skeptics, however, if it doesn’t do it all, it is not intelligent.

Most, if not all, of the components identified in the modeling exercise presented in this section (see Model Components in the next section) will need to be present in a truly neuromorphic computer model. If a machine with parts like these can imitate cognitive behavior, will it be considered intelligent? Probably not, at least not as long as “artificial intelligence” is a moving target (perhaps one that even dodges when a technophobe approaches).

Here is a question for you to think through: On the cosmic scale of sentience, with abiotic things such as rocks at the bottom, animals and humans in the middle, and divine beings at the top, will man-made machines ever be able to totally surpass man?

Caveat Emptor

As I said in one of my introductory posts, I do not want to be guilty of greedy reductionism. This is a common phenomenon among impatient researchers who would like to arrive at conclusions before testing all the premises or even ensuring that they know all the applicable constraints from which to establish premises. To avoid this I will clarify my intent:

  1. Eclectic Components of Cognitive ScienceI do not know enough to attempt to explain the brain, cognition or consciousness, just to model some phenomena using computers
  2. I do not know enough about learning to claim anything, just enough to mimic learning processes using computer software
  3. I do not know any language well enough to claim to be an expert in that language — I’m just a trained observer of language phenomena
  4. I do not know enough about computers to solve every automation problem, but I know enough to find a solution given the right constraints

It may be the case that, if we design a machine capable of understanding human intent, we will learn things we didn’t know about human capabilities. Or maybe not. If this blog accomplishes anything, it will be to show people from one of the disciplines assembled for this work, that there are important findings in other academic areas that can contribute to building a better mousetrap, so to speak, when the mouse in the crosshairs (unfortunate mixed metaphor) is understanding human intent.

Click below to look in each Understanding Context section


 

 

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.