Welcome to the A.L.I.C.E. AI Foundation

Promoting the adoption and development of Alicebot and AIML free software.

You want a revolution?

Dr. Richard S. Wallace

Dr. Richard S. Wallace

If you are looking for a revolution in IT, you've come to the right place.

There are several ways in which A.L.I.C.E. and AIML defy the dominant computer science paradigms.

For one thing, the ways in which humans and computers normally communicate are very different. Humans tend to spend a lot of time on chit chat and informal dialogue with little or no purpose. Computers are known for giving precise, true, and logical answers. The rate of information exchange in most human dialogue is very low, no more than 1Kbit per second, but computer communication is much faster. Alicebot/AIML is an attempt to bridge this divide.

We have dispensed with much of the conventional wisdom from structured programming, too. Where less code is usually good, more AIML is usually better. Filling up RAM with a lot of code that is seldom activated is a good idea in AIML. Allowing a novice programmer to write a large and unwieldy program in AIML is also, unconventionally, good practice.

Not to mention, the basic mimimalist approach of AIML is borrowed from ELIZA, a design that was largely abandoned by the research establishment, indeed even derided as a "toy". Yet the success of A.L.I.C.E. and AIML if anything puts the research establishment to shame: for their years of research and millions of spending, they have achieved nothing close to the performace of A.L.I.C.E. in natural language processing. Even the success stories of AI research, such as the chess-playing supercomputers, required far higher expenditures than A.L.I.C.E. and AIML.

The conventional model of an information service assumes that a computer will always provide accurate replies promptly. People have asked, what is the difference between Ask Jeeves and A.L.I.C.E.? Jeeves is designed so that the client will ideally ask one question, and then immediately receive an accurate reply including a hyperlink to another site. After just one transaction, the client clicks on the link and leaves the Ask.com site. A.L.I.C.E., on the other hand, is designed to keep the client talking as long as possible, without necessarily providing any useful information along the way. The longer average conversation lengths measured over the years have in fact been a measure of A.L.I.C.E.'s progress.

Strange as it seems, free software is still not considered a mainstream methodology for either research or software development. As big and successful as Linux is, the majority of the world has not yet bought the free software argument. For many people, the whole debate over software intellectual property remains obscure. Add artificial intelligence to the mix, and you have something truly unique. There are a few other free software AI projects out there, but none has had the impact of Alicebot/AIML. This year was our first appearance at Linuxworld, a first step in finding common ground between the AI and free software communities.

Finally, many in the Alicebot/AIML community share a common vision of the future of our technology. Alicebot is perceived as a "missing piece" of the puzzle that combines speech recognition, natural language understanding, and voice synthesis. The vision is the talking Star Trek/HAL-style computer of the future, the invisible piece of hardware that responds to voice commands. In that vision, there is no keyboard, display, mouse, or graphical user interface. If that vision comes true, the impact of AIML will be ubiquitous.