AI
I consider myself to be privileged in some ways as I have been around at a time which has seen computers become mainstream and watched technology change the home and the world from the days where I played tennis on my ZX81 all the way to what we see today.
Now I am no guru at all. I did study AI at university back in 1994. At the time the technology did not match the dreams. But times have moved on. If you look back what you can see is a few trends. All of the trends have one meta trend in common. Abstraction.
Hardware performance grew and grew until you did not have to programme in machine code. The hardware coped with more and more of the functions you needed (from basic adders to the graphics cards which now do most of the complex work for you).
OS went the same way. The OS was fairly basic at first and over time (and enabled by improvements at the hardware level) the OS has abstracted away from hardware control to something like Apple's tiger where the experience of the user is one that requires almost no technical knowledge at all. Same for the UI. You no longer operate a computer, you carry out tasks. I am not sure if I could pick a time when this occurred but it was relatively recently. It is the reason why Microsoft will ultimately go down if they do not catch up with this trend.
The internet went the same way. From hard coded HTML all the way up to modern content management and various SAAS offerings like Facebook, Google etc etc.
So the trend will continue but there is only so much more that can be done. Nobody operates a computer any more.
What is needed to progress now is real intelligence. We now have platforms that basically allow the human to carryout the tasks they need to do but the common thing is that we are still alone. The computer adds little to help us. What is needed in the next leap forward is genuine intelligence. Proactive help (not a little paperclip with a search tool linking to a forum). This is hard. The functionality in most apps is now well beyond the ability of the user. The future of computers is not for AI to carry out the simple tasks but to get the computer to intelligently carry out the enhancements that you want.
What do I mean? Well, take a simple office based process. I want to create a doc and send it out. Today, even if I try and use a template, I will be wrestling with the package and then have to try and distribute to a mail merge list via email. Its hard. But I know what the outcome I want is. So intelligent systems will need to understand the outcome and carryout/support your attempts to achieve that outcome.
The trouble with AI at the moment is that it is too focused on human replication. Rather than trying to replicate human ability more focus should be put on making the computer systems intelligent.
Now I am no guru at all. I did study AI at university back in 1994. At the time the technology did not match the dreams. But times have moved on. If you look back what you can see is a few trends. All of the trends have one meta trend in common. Abstraction.
Hardware performance grew and grew until you did not have to programme in machine code. The hardware coped with more and more of the functions you needed (from basic adders to the graphics cards which now do most of the complex work for you).
OS went the same way. The OS was fairly basic at first and over time (and enabled by improvements at the hardware level) the OS has abstracted away from hardware control to something like Apple's tiger where the experience of the user is one that requires almost no technical knowledge at all. Same for the UI. You no longer operate a computer, you carry out tasks. I am not sure if I could pick a time when this occurred but it was relatively recently. It is the reason why Microsoft will ultimately go down if they do not catch up with this trend.
The internet went the same way. From hard coded HTML all the way up to modern content management and various SAAS offerings like Facebook, Google etc etc.
So the trend will continue but there is only so much more that can be done. Nobody operates a computer any more.
What is needed to progress now is real intelligence. We now have platforms that basically allow the human to carryout the tasks they need to do but the common thing is that we are still alone. The computer adds little to help us. What is needed in the next leap forward is genuine intelligence. Proactive help (not a little paperclip with a search tool linking to a forum). This is hard. The functionality in most apps is now well beyond the ability of the user. The future of computers is not for AI to carry out the simple tasks but to get the computer to intelligently carry out the enhancements that you want.
What do I mean? Well, take a simple office based process. I want to create a doc and send it out. Today, even if I try and use a template, I will be wrestling with the package and then have to try and distribute to a mail merge list via email. Its hard. But I know what the outcome I want is. So intelligent systems will need to understand the outcome and carryout/support your attempts to achieve that outcome.
The trouble with AI at the moment is that it is too focused on human replication. Rather than trying to replicate human ability more focus should be put on making the computer systems intelligent.
posted by So long and thanks for all the fish at
12:30 pm
0 Comments:
Post a Comment
Subscribe to Post Comments [Atom]
<< Home