Learning Programming In The Era Of Artificial Intelligence

Learning Programming In The Era Of Artificial Intelligence

The biggest problem in computer programming is where to start, and the best answer is: by making your computer talk, and maybe think.

As of 10 years ago, we have a tool to automate the process of programming, and the tool is really great at a lot of things – and that is also its weakness.

It is large, it is slow, it demands large computers, and it does a lot more things than you need.


There is no better way to get into programming, than to use the big AI to make a better AI.

Smaller AI, faster AI, one that runs on a small computer, one that does exactly what you need it to do.


But, you also have an opportunity today to not just learn programming, but also have your AI make a tiny little company that makes applications.

And preferably applications, that when subscribed to or purchased, make people money, this way, you make money and they make money.

Kind of fair, and kind of square.


So the intelligence you should be learning programming with, should make applications, preferably web applications on a web desktop.

Because this way it makes sense to log into their desktop, and the first app they will see, is you app store, which makes more sense.

You already login to your laptop, you can see that you can have multiple users, and you would just use that functionality on a Linux server.

User home directories, user files.


To start making your own AI, think about text adventure games, but not for you, for your little low powered AI.

Look up an old game called Lemmings, or you can imagine a Starship away team.

You will equip these agents with SKILLS that make them specialists, and then send them into a virtual world of text description.

But not into the traditional MultiUser Dungeon lobby, but an empty window on a Desktop.

Your application layout agent can split the window into a file pane on one side, and a properties pane on the other.

You can see how you don’t need much power to build layouts, provided that you have a web desktop with a window ready to go.

And depending on what you tell your most powerful AI to build, your component agent, can for example go inside the center pane.

And place a control there, for displaying images.


Your big AI will design the worlds and maps, and agent minds, their SKILLS and maybe concept maps that react to world environments.

All of your agents will be AI but some the good old fashion kind, often used in car diagnostic computers.

Programs like yeoman already demonstrate building code out of code, and your AI would just make it more lively, more interesting.

At the end of the day you may still need a strong AI, but it would be one that costs cents, not dollars.

And again the reason why you would want your own AI is that, it can be faster.

You start building your applications with a preset environment, knowledgeable agents of all shapes and sizes.

You AI will be assembling your applications out of wiser and sturdier pieces, and that is a lot better than a chat window.

Before you know you, you will have many such custom AI experiments, and eventually an AI that for a time, will be the best at what it does.