Vision

In developing the Golden Gallery project, our goal is to make it easier to learn about, experiment with, and write programs that think the way humans do. We have picked three excellent AI programs which were written in the 1960s - 1970s and have rewritten them in modern programming languages. We have made them run on the web, written detailed manuals to explain how they work in plain language, and released them under a free software license which allows you to download and modify them yourself however you like.

On this website, we share the results of our work. We hope that these programs will be useful to many different people:

• Some may be inspired by our live demonstrations, which show the possibilities of programs that are not only smart, but that also model the way we humans think.
• Some may want to learn more about how these programs work behind the scenes — how do you examine your own thinking processes, and how do you replicate those processes in code?
• Some may want to download a copy of the programs for themselves, to test it out and see what else the programs can be made to do.
• And some may want to learn a new skill: a program is often a very detailed description of how to solve a particular kind of problem, and so they make excellent tutors. Here, for example, if you are learning how to solve calculus problems, or you want to understand more about analogies, you may learn a great deal by watching these programs explain how they think.

Demos

Two of our programs are written in Javascript and run directly in the browser. You can see a live demonstration of SAINT here, and a live demonstration of ANALOGY here.

Code

Download the code for our projects here. Each project is distributed as a TAR file.

• SAINT makes educated guesses when solving calculus problems.
• ANALOGY uses analogical reasoning to solve visual puzzles.
• HACKER learns to build block towers by making and debugging plans.

We license all of our code under the GPL 2.0+ license. This means that you are free to download, modify, and re-share our code however you like, as long as you share it in the same way.

Docs

All our documentation is literate, meaning that it interweaves the source code and explanations of how the code works.

• Manual for SAINT
• Manual for ANALOGY
• Manual for HACKER

We also provide links to essays describing the programs on which our programs are based^*.

• Jim Slagle's brief synopsis of the SAINT program.
  A heuristic program that solves symbolic integration problems in freshman Calculus
• Tom Evans's brief synopsis of the ANALOGY program.
  A heuristic program to solve geometric analogy problems
• Gerry Sussman's PhD dissertation describing HACKER.
  A computational model of skill acquisition

Contact

We would love to hear from you! Write to golden-gallery @ logical.ai or to dxh @ mit.edu.

* These essays are not covered by the Golden Gallery's Creative Commons license; you may have to ask permission from their copyright holder to use them.