What's next for intelligent computing

Computers may have been invented to automate our lives and make them easier, but in many ways they are like bad employees. They don't really understand what we want, aren't able to use their own initiative, and can often require so much guidance that it would be easier to do the job ourselves. Is there such a thing as intelligent computing, and could it finally deliver a system that just gets the job done without any fuss? There are some potential advances in intelligent computing. Here are three:
##CONTINUE##
Computers that know what you want

Generally, computers are like uneducated, disinterested retail assistants. If you ask them for something, they will have no idea what it actually means, but may try to find it by searching for keywords. Compare that to a retail assistant who has a passion for what they are selling. You ask them for something, and they instantly know what you are likely to use it for, what other products it is likely to be used with, and whether there are any competing products that are better. Good retail assistants have that kind of expertise in their heads. You can replicate that in a computer system using semantic tagging technology. This encodes that information invisibly into the data, so that the computer can read it. The computer can then understand what the data means, and can use it to return more meaningful results.

In practice, that might mean that when you search for a company online, instead of simply returning all of the text results containing that company name, the computer automatically retrieves information about key executives, customers, suppliers, and products. The computer will almost seem to be an expert in its own right, simply because the data has been properly tagged with semantic information that makes it more meaningful.

Further reading: Semantic web Wikipedia page

Computers that know what you're doing

This is one of the tougher nuts to crack when it comes to intelligent computing. Think about the way that you interact with your computer today. Does it work with you politely, or does it interrupt what you are doing at inopportune moments? When I am past deadline trying to get an article finished (like this one), I don't want my computer to suddenly tell me that it has updated its software and is about to restart. Similarly, I might not want my calendar software to remind me just at that moment that I have a meeting in 10 minutes, if I am trying to finish a spreadsheet before I leave the office.

In fact, if computers were really smart, they could go further than not interrupting you themselves. They would be able to understand when you on the phone, and who you were talking to. Then, if another call on instant message arrived from someone who you didn't consider to be that important, the computer could hold on to that message until it thought you were ready to receive it. If, on the other hand, the boss was trying to get hold of you, the computer might decide that the blog entry you were currently working on could wait for a couple of minutes while you took her call.

This requires the computer to marshal a lot of information about your contacts, your relationship with them, and your current activity. Most of this information is actually available now -- but it requires some smart software design to ensure the right balance between a computer that is too needy, and a computer that blocks everything for fear of interrupting you unduly. It also requires someone to integrate the computer with the phone system, and to pull that altogether with your instant messaging software, and with your back end scheduling service. That is no mean feat.

Further reading: Polite Computing article, Technology Review

Computers that know where you are

This, at least, is something that we are already seeing. Right now, location-based computing is in its infancy. Products like the iPhone make it possible for you to know where you are, and to look for local attractions and resources, or to plot directions to your meeting. But there is much more that phones and the back-end computers that support them could do. For example, a software application that knew you had been searching for a particular pair of shoes online might know that the shop you are passing has those shoes in its inventory. It could alert you that your prize pair of shoes was just a few yards away. Perhaps the retailer could even give you an automatic discount to lure you into the store.

Again, the information to make that work is already available, but there are two challenges that stop this happening today.

The first is integration. Most software applications still have a hard time integrating with each other to share information about you. Instead, they hold that data in 'silos', making it difficult for even single companies to have a single view of a particular customer, let alone multiple companies partnering together.

The other problem is privacy. There are limitations on how companies can share data about us. That is what makes it harder for a Big Brother-style society to develop, in which companies can know everything about you -- including the information that you didn't explicitly give them. Many people would argue that privacy is not a problem, it is a right. Intelligent computing may seem like a fabulous idea, but occasionally, it is smarter to keep our computers just a little bit dumb.

Further reading: Presentation on privacy and location-based services

-----------------------------
BY Pamela Heaven
Source：FP POSTED

© 2009 The National Post Company. All rights reserved.