Categories : Commerce/Financial Communication Computing Data/Information/Knowledge Governance Human-Machine Interface Information Filtering - Fact Checking Integrity Medicine (non-Malthusian) Teaching People - What and How Tools for Society Uncategorized

Reviewed By lobobear - Rating : 5.0
 

In the previous post it became clear that the difference between humans and machines is the speed of processing data and information. A simple data point is really very simple and we humans can handle it very easily even if we do not understand the potential knowledge it conveys. If you go to the grocery store and see an item for sale for one dollar, it clearly tells you how much money you need to pay for it. You fill your cart with a dozen items, but you do not know the total precise bill until you check out and the machine adds if up and adds sales tax (ok, gross receipts tax) when applicable. While you could carry a pad of paper with you or use your smart phone to maintain a running total, we rely on the machine to figure it out. In fact, the store wants the machine to do it rather than the checkout person.

 

Now consider what you want to watch on TV tonight. In the 1950s, where I grew up, there were three broadcast channels one for each network (Fox was not yet into TV broadcasting). During the “prime time” hours of viewing each channel had three to six shows of one or one half hour in length, 9 to 18 shows every evening, total. And the schedule was consistent every week, a small data set with a maximum of 126 shows per week. As an individual, it was very easy to make a selection – I watched what my mom and dad wanted to watch. Flash forward to today and examine the scheduling. Three hundred plus channels (probably many more) plus streaming services where thousands of programs or media content is available at any time. How do you make a choice and how do you find it?

 

Change your perspective and look at the problem as a media provider. The provider is offering a product(s) for sale and wants to maximize resources based on users of the service. If the service is paid for by advertisers, the quantity of viewership influences the fees and revenues. If the service is paid for by customers, viewership directly affects the revenues received. In the advertising model, the provider really does not care about what you, as an individual, want to watch. Rather it is all about the numbers and the demographics. But if it is a streaming or subscription service being paid for by the customer with her or her personal money, the customer has to find what he or she wants or the customer departs. Are the customers individuals or advertisers?

 

When my wife and I had a store here in Los Alamos, we tried to remember customers and what they wanted to buy or consider buying. Our slow computers called brains tried to remember names, faces and preferences or we recorded the data on our hard drive called a notebook. Of course, accessing the “hard drive” was a slow process. We were slow. And we, at best, had only a thousand or so items for sale. Further, when it came time for restocking, we used sales data and our vague perceptions of customer satisfaction.

 

Change the product to streaming entertainment and a service such as Netflix. Currently Netflix offers well over 5,000 viewing options available to all customers at any time. But how do you, as a customer, sort through and examine over 5,000 products, growing and changing continuously. While our store had, at best, 1,000 or so products, it is the same problem that faced our customers. Actually, it was our problem. Since we had only two or three customers at a time, we used our intelligence and slow CPUs to help them. But what if you had millions of customers simultaneously? We would have needed help from a machine, a computer. And the machine would have had to have been intelligent in the manner that we knew our customers and reasoned about what to show or offer them. It would have been a machine, programmed with algorithms written by humans. It would have been a machine with manmade intelligence – artificial.

 

When you first start using a streaming service, it offers you a standard presentation of material available. When a new potential customer entered our store, we asked him or her what he or she was looking for: wall art, jewelry, pottery, or something else. The customer responded and made a choice. We noted it. For the streaming service, assume the new user opts for an action movie. The software managing the service notes that customer XYZ chose and action movie and alters the algorithms for customer XYZ to show a preference for action movies. In a sense, the software remembers customer XYZ’s preferences and reprograms itself. The software was created to learn about the customer and adapt to customer wants. This is intelligence. But it was created by human programmers – it is artificial.

 

But what if the programmer got it wrong? The program, the machine, is making real time choices and decisions about presenting the user, the human being, with information upon which to base a human decision, to make a choice. And the machine can “reprogram” itself. If the program makes a mistake, what is the cost of the error? It probably is not very much when you have millions of customer. But what is the consequence if the machine, the program, is to reduce natural gas consumption in your house? Winter is coming.

Print Friendly, PDF & Email
 Posted on : January 14, 2019
Tags :

Leave a Comment

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

You must beLogged in to post a comment.