Overview Index

Computer intelligence versus Human intelligence

Intelligent systems (both natural and artificial) have several key features. Some intelligence features are more developed in a human’s brain; other intelligence features are more developed in modern computers.

Name of intelligence feature	Who has the advantage	Comments about comparison
Experimental learning	Human	Currently computers are not able to general experimenting. Computers are able to make some specific (narrow) experimentation though.
Direct gathering information	Computer	Modern computers are very strong in gathering information. Search engines and particularly Google is the best example
Decision making ability to achieve goals	Human	Currently computers are not able to make good decisions in “general” environment
Hardware processing power	Computer	Processing power of modern computers is tremendous (several billion operations per second)
Hardware memory storage	Computer	HDD memory storage of modern computers is huge
Information retrieval speed	Computer	Data retrieval speed of modern computers is ~1000 times faster that human’s ability. Examples of high-speed data retrieval systems: RDBMS, Google
Information retrieval depths	It’s not clear	Both humans and computers have limited ability in “deep” informational retrieval.
Information breadth	Computer	Practically every internet search engine beats human’s in the breadth of stored and available information
Information retrieval relevancy	Human	Usually human’s brain retrieves more relevant information than computer program. But advantage of humans disappears every year.
Ability to find and establish correlation between concepts	Human	Currently computers are not able to establish correlation between general concepts
Ability to derive concepts from other concepts	Human	Usually computers are not able to derive concepts from other concepts
Consistent system of super goals	Human	Humans have highly developed system of super goals (“avoid pain”, “avoid hunger”, sexuality, “desire to talk” …). Super goals implementation in modern computers is very limited.

Features of intelligence

Learning ability

Learning ability consists of two parts:

1. Experimental learning (acquire knowledge through experimenting).
2. Direct gathering of information (from books, web-sites, conversation).

Ability to achieve goals by decision making

Hardware power

1. Processing power
2. Memory storage

Information retrieval from the internal memory ability

Three characteristics are important for retrieval ability

1. Retrieval speed. (That is how quick can system find necessary information and make this information available for further processing).
2. Retrieval depth and broadness. (That is how deep information searching process is. How many related concepts system selects during retrieval.)
3. Retrieval accuracy.
   Retrieval accuracy shows how accurate retrieval process is. What the “useful concepts/total concepts” retrieval ratio is.

Examples:

Let’s imagine that system tries to find answer to the question “How to talk to Mary?”

• Narrow retrieval system will return answer: “Make a phone call to Mary”
• Broader retrieval system will return answers: “Make a phone call to Mary”, “Email to Mary”, “Meet Mary on the next party”, “Come to Mary”…
• In-depth retrieval system will return answers like: “Ask Susan to call to Mary (because Susan is a friend of mine and a friend of Mary, so Susan will easily convince Mary to talk with me)”.
• Accurate retrieval system will return answers like: “Make a phone call to Mary”, “Mary’s phone number is 123-456-7890”.
• Inaccurate retrieval system will return answers like: “Mary is a woman”, “Talk is the pleasure”, “’How’ is a word for the questions”, “Talk to Mary if you want”…
• Slow retrieval system will return answer in several hours.
• Fast retrieval system will return answer in one second.

Deliberation ability

Deliberation ability is:

- Ability to find and establish correlation between previously not related concepts.

- Ability to derive concepts from other concepts.

Example of concept derivation:

Let Intelligent System knows concepts “mother”, “father”, “brother”, “sister”. After deliberation process Intelligent System unites these concepts under concept “family”.

Example of establishing correlation between different concepts:

Let Intelligent System knows that “have a job” is a cause to “salary”. Let Intelligent System also knows that “salary” causes “ability to pay for food”. As a result of deliberation process Intelligent System would conclude that “have a job” is a cause to “ability to pay for food”.

Consistent system of super goals

Intelligent System should have convenient/understandable/easy-to-use set of super goals that direct system self-development.