by Darius Kazemi, June 1 2019
Artificial intelligence online
The technical content
The Artificial Intelligence Group at Stanford Research Institute expects to connect to the ARPANET in the summer of 1971 via their PDP-10. They plan to use the IMP currently connected to SRI ARC.
Mid-July 1971 is their “most optimistic estimate”.
Once online, they plan to support several simultaneous remote users during the business day. In particular they are excited to provide access to
- QA3.6, a question-answering system (hence QA) which is “a first-order resolution theorem prover”
- STRIPS, the Stanford Research Institute Problem Solver
It's tough to find information on QA3, but it's described in this paper from 1969 or 1970 as
programmed in LISP on the SDS 940 computer, operating in the time-sharing mode. The user works at a teletype, entering statements and questions, and receiving replies.
A user session might look like this:
[User states that Smith is a man]
[User asks is Smith a man]
[User states that all men are animals]
[User asks “Is there some “y” such that “y” is an animal? If so, show me”]
ANSWER: YES, y = Smith
So the basic idea is: you input formal logical statements, and then you can make queries based on those formal statements.
More on STRIPS at Wikipedia. It seems to offer a similar kind of service.
B. Michael Wilber worked on many projects at SRI's Artificial Intelligence Group, including the famous robot SHAKEY.
This was, incidentally, right before ARPA made big funding cuts in artificial intelligence that led to what's known as the first AI winter.
How to follow this blog
I'm Darius Kazemi. I'm an independent technologist and artist. I do a lot of work on the decentralized web with ActivityPub, including a Node.js reference implementation, an RSS-to-ActivityPub converter, and a fork of Mastodon, called Hometown. You can support my work via my Patreon.