A repository to collect papers and programs of historical interest to AI. Mostly gathered while reading Pamela McCorduck's Machines Who Think
- Newell (1965) - Limitations of the Current Stock of Ideas about Problem Solving
- Minsky (1961) - Steps Toward Artificial Intelligence
- McCarthy & Hayes (1969) - Some Philosophical Problems from the Standpoint of Artificial Intelligence
Turing's seminal AI work... A nice read with a curious section about ESP!
Claude Shannon's original 1950 paper...
An article in Scientific American. Super readable... Scientific American Article
McCarthy, Minsky, Rochester, Shannon (1955) - A Proposal for the Dartmouth Summer Research Project on Artificial Intelligence
Proved 38 of the first 52 theorems in Russell & Whitehead's Principia Mathematica and even found a new and better proof for one.
Source Code has not been transcribed and doesn't appear to be easily available.
Early speculative paper on something McCarthy called the "Advice Taker". Possibly the first paper mentioning "Common Sense" in terms of AI. Also contains an interesting discussion with Bar-Hillel and Selfridge as an appendix.
A sort of summary/review article covering the approaches and advances in AI at that time
Very generic extension of their Logic Theory Machine, separating problem domain from the process of problem solving and introducing concepts like Means End Analysis, Planning, Goals, Sub-Goals and Differences.
A more readable paper with emphasis on the comparison between how GPS works versus human subjects...
Newell & Simon (1963) - GPS - A Program That Simulates Human Thought
This program could answer questions about baseball (for a specific year) specified in plain english using syntactic analysis...
Lindsay (1963) - Inferential Memory as the Basis of Machines Which Understand Natural Language (SAD SAM)
Hardly mentioned in the paper, which is more of an early analysis (an excellent one) of the difficulties of natural language processing, is SAD SAM which was the name of the program which could examine kinship relations.
STUDENT was a LISP program that could accept mathematical puzzles in a limited set of English, and solve them.
ANALOGY was a LISP program designed to solve simple Geometric Analogy problems like '''Figure A is to Figure B as Figure C is to...'''
This is magnificent. Short, concise survey of the known methods of solving problems, and an incisive look at a problem we don't know how we could go about solving/proving, and how we as humans come across the solution from 'out of nowhere'... The example problem is the 'mutilated chess board' checker problem.
Live version can be found here: https://www.masswerk.at/elizabot/
- Javascript version: https://github.com/brandongmwong/elizabot-js
- Python version: https://github.com/wadetb/eliza
Much more detail on all kinds of ELIZAs here:
McCarthy & Hayes (1969) - Some Philosophical Problems from the Standpoint of Artificial Intelligence
Detailed and philosophical approach to the problems of AI, trying to develop a mathematical logic approach to it. Discussion of Modal logics of the time as possibly useful.
A couple of modern/ports are available on github:
Very early natural language understanding system written in Lisp by Terry Winograd. The system manipulates a simple block world by accepting commands in English. The original paper/thesis is available here:
http://hci.stanford.edu/winograd/shrdlu/AITR-235.pdf
Early speech recognition within a domain (Chess Moves)
A modern python port of the analogy making program Copycat can be found here: