ELIZA The Earliest Chatbot

I was nineteen when I first talked to a chatbot. It was 1991, ELIZA was on the curriculum, and I had gone to the University of Sussex to find out how ‘everything’ worked.

I’d started a Philosophy degree in the School of Cognitive and Computing Sciences (COGS). I chose it because I wanted to understand the nature of things. Philosophy seemed the closest degree.

I lasted one term, as I wasn’t ready for university life, but in that short time I studied AI, and ELIZA was part of it. We worked through Descartes and the nature of thinking and understanding, then sat down to program ELIZA, that appeared to be mirroring ‘intelligence’.

What I didn’t know was that COGS had been founded four years earlier by Professor Margaret Boden, who built the world’s first academic programme in cognitive science and became its founding Dean. By the nineties it was the leading centre for AI in Europe. She spent her career on a single question – whether a machine can genuinely think and create, or only appear to.

What was ELIZA?

ELIZA was built in the 1960s by Joseph Weizenbaum at MIT. Its best-known script, DOCTOR, published in 1966, simulated a Rogerian therapist – picking up key words in whatever you typed and reflecting them back as a question.

As I was taught it, one of the proposed uses was a sort of business therapist for executives. A director would sit down in the morning and ELIZA would open with something like “good morning, how was your weekend?”

It’s a detail that’s stuck with me for thirty-five years. Even designed as an early AI Chatbot for the office, the whole appeal was the illusion of being asked ‘how you were doing’.

People knew it was a program and confided in it anyway. That phenomenon has a name now – the ELIZA effect. Weizenbaum found it alarming enough that his 1976 book, Computer Power and Human Reason, became a sustained argument against trusting computers in the places where human judgement really matters.

The man who built the first chatbot warned people not to trust it even back then.

The lost source code

This year marks the sixtieth anniversary of ELIZA’s public debut, and a new book was published: Inventing ELIZA: How the First Chatbot Shaped the Future of AI, the result of over four years of archival work by eight scholars across the University of Sussex, MIT, Oxford, Stanford and USC, built around the original ELIZA source code they recovered from the MIT archives after it had been lost for over half a century.

One of those eight scholars is Professor David M. Berry, Professor of Digital Humanities at the University of Sussex – the same university where I sat in a School of Cognitive Sciences classroom in 1991, working through ELIZA as coursework.

The University of Sussex posted about the book on its own channels last month. Seeing that post, at the same time I was writing this article was a strange synchronicity, bringing me back full circle to my former University.

Science reviewed the book in its 18 June issue, calling it “a masterful contribution to the emerging discipline of software studies.” A monograph on 1960s chatbot code doesn’t usually get that kind of attention from one of the world’s leading scientific journals – a fair sign that ELIZA’s relevance to how we think about AI hasn’t faded with age.

Ecology taught me to think in systems

I did go back to Sussex. Not to philosophy but to ecology, still on a mission to find out how everything is connected.

I graduated in 2000 with a BSc in Ecology and Conservation, and that degree gave me the answer the first one hadn’t. Not an answer about thinking. An answer about connection.

Ecology is the study of the interconnectedness of life, and that’s how I’ve looked at everything since, including the digital work I do now.

A digital presence is a system too. Website, emails, social, the search results people arrive through – none of it works in isolation, and the whole is only ever as healthy as the connections between its parts.

Ecologist and digital strategist sounds like an odd pairing until you notice they’re the same discipline – systems thinking.

Going back a third time

This year I return for the third time. I’ve enrolled on the University of Sussex’s AI Literacy for Communicators course. Since I use AI tools daily, I need to gain a deeper understanding of how they work with some solid academic training.

Thirty-five years ago ELIZA asked me how something made me feel. Now I’m asking the question – how do I feel about using AI? Which has sent me back on another ethical quest….