Demystifying AI in Everyday Life
By ai-depot | August 16, 2002
Shows how AI can provide computers with a set of rules for decisions-making, in a fashion similar to humans. Such AI systems gather information, filter it discarding irrelevant details, and produce an answer suited to solving the problem at hand.
Written by Nick Loadholtes.
Demystifying AI in Everyday Life
Abstract
The goal of the field of Artificial Intelligence (AI) is to produce systems (be they robots, computer programs, user interfaces, etc.) that utilize behaviors and decision making skills that are similar to ones that humans (and other living creatures) use. In essence, to create or mimic intelligent behavior in non-sentient entities. The wealth of knowledge that has been gathered in researching this topic reveals much about logic, perception, and common sense. This essay intends to show that the principals behind AI are not out of reach to the common persons; in fact, these principals can be utilized in many common situations to highlight the basics of human thinking.
Real Life
Artificial Intelligence (AI) is a field of study in Computer Science that is becoming more relevant to our lives everyday for many reasons. Computers are wonderful tools, they are able to do math faster and more accurately than we can, they help us to communicate, they even help us organize our businesses and lives. However, for all of the great things that computers do, there is one thing that that they can’t do very well: Think.
The goal of AI is to produce machines (computers, toasters oven, cell phones, etc.) that can “mimic” human intelligence. For example, if your toaster oven detects that the toast is burning, it should turn off the heat. For a human, this is an obvious thing to do. If you let the toast burn, it will probably start to smoke and set off the smoke detector (not to mention the horrible smell it will produce). For the toaster, and the computer that controls it, this is not so obvious and thus it will burn the toast.
If however, the toaster “knew” some things that we humans take for granted, it could prevent this happening. This is the goal of Artificial Intelligence, to allow machines act in an intelligent way, to do what we humans would in that situation. This sounds like a very difficult thing to do, and in some ways it is, but the basic principals are very easy to understand.
Common Sense
To do this, researchers and scientists over the years have worked to distill the mechanics of human knowledge and put it into a form that computers can understand. Computers are very good at understanding yes-or-no questions (also known as true-or-false statements). So by studying the thought processes and decision making processes of people, a set rules have been ‘discovered’ that are very simple in nature and easy to communicate to a computer.
A person makes a decision by doing these things: Gathering information (through the sense, by remembering, etc.), deciding what information is important, then acting on that information. This simple set of directions is the basis for most AI systems.
Consider this: For an AI system to be successful, it must look at the data it is presented and determine what is important (some times referred to as the ’signal’) and what is not important (referred to as the ‘noise’). Then using this data, or signal, it will ask a series of questions to try to help it achieve its goal. Using the burning toast example from above, the goal would be to toast a piece of bread and the data presented to the toaster may be this:
- How long the toast has been toasting
- The temperature setting of the toaster
- The model number of the toaster
A person can look at this set of information and very quickly determine that c is not important to us (because it doesn’t tell us if the toast is in danger of burning). This is the ‘noise’ and a and b are the ’signal’. In an AI system there must have a set of rules the will help it discard the noise and focus on the signal.
In this case, there would be a rule that answers the question “Is the toast in danger of burning?” This rule would then have the computer in the toaster oven check the temperature setting and how long the toaster has been toasting. From there, it is easy to figure out if the toast has been in there too long. (For example, if the temperature is set to 300 degrees, and the toast has been in the over for 10 minutes, the chances are very good that it is burned). If it is ‘true’ that the toast is in danger of burning, there is another rule the toaster will go to which says, “This is what to do if the toast is in danger of burning.”
Cognitive Processes
This approach to making computers smarter has an interesting side affect: By studying how we humans make our decisions, we not only are able to make our computers smarter, we are able to better understand how we think.
This is an important idea because we now have a rudimentary blueprint of how humans are able to make decisions. By studying how AI systems deal with decision making, we can take and apply the lessons learned to our own everyday lives. In Philosophy there is a subject called Critical Thinking that details the same process of decision making. However, using the perspective of an AI system (and how you would explain something in very basic terms to it) can sometimes make these processes easier to understand.
Other Examples
This, of course, is just one example of how an AI system works; there are many different approaches and techniques. However, they all share the same basic idea: Gather information, then see if the information can help answer a question, then perform an action based on the information.
And this is not very different from how people work. Before a person crosses a street, they will look both ways to see if there are any cars coming (this is gathering information), and then based on whether they see any cars coming they will decided to cross the road. The only real difference between a human decision and a computer decision is that a human is more willing to “bend the rules”, i.e. if they see cars but feel that they are far enough away, a human is willing to risk it and run across the street.
Computers, obviously, are not willing to bend the rules. If your bank account has $10 in it and you attempt to withdraw $20 from the Automatic Teller Machine (ATM), it will not give you any money. This is a perfect example of a way in which AI systems can be improved; to understand what to do in these odd situations that there is no rule for (or the rule that exists is not sufficient to cover the situation).
This gray area is one where AI systems will eventually learn to handle. In the example above, the ATM could have told you that you only have enough money in your account for it to give you $10. Creating that type of intelligence is possible today, but it is not commonplace yet. But as AI systems learn become more proficient at making decisions based on implied information (implied in this case meaning that although the ATM could not give the exact dollar amount that was requested, it could have given some smaller dollar amount), this “intelligent” behavior will become more of an everyday experience.
In summary, Artificial Intelligence provides computers with a set of rules that it can use to make decisions that would be similar to what a person would do in the same situation. An AI system will gather information then take that information and try to answer a question. In the process of doing this the AI system will filter through the information and discard information that isn’t relevant to the question it is trying to answer. This is a very similar process to how people make everyday decisions: gather information, evaluate it, and then use it to help achieve a goal.
Written by Nick Loadholtes.
Tags: none
Category: essay |
