From Tantra to Apple II

He was a hardware hacker who built his computers from parts that he found at Radio Shack and who knows where. His curiosity led him to strange places, like my Tantric meditation workshop. He must have enjoyed it because the day after, unsolicited, he offered me one of his Apple II computers for free.

The year was 1982, and I was freshly transplanted from Paris to Berkeley. Teaching Tantra was my way to deepen my understanding of sexuality as worship, drawing on my experience of a half-year retreat in an ashram in India, where I shifted out from my academic career for the much more exciting inner journey, as many Western intellectuals did in the ‘70s. I thought that I will never return to academia but was still too young to understand “never say never.”

During my Paris and Berkeley years, I had a frequent urge to write poetry, for which I used an old Remington typewriter, and I didn’t see why I would need a computer. When my hacker friend showed me that to replace a word, I didn’t have to pull out the paper from the typewriter carriage and white it out (obliterate the mistake with white correction fluid) before typing the new word over it; I could correct it with just a few clicks in the wordprocessor software. I was sold.

At that moment, I hadn’t yet fathomed the new worlds waiting for me to discover thanks to a modem, the little box that connected my Apple II with the phone line and through that with friends and strangers in remote lands.

Diving into EIES

The first of those unknown worlds was the Electronic Information Exchange System, created by the same Murray Turoff, whose Network Nation: Human Communication Via Computer, co-authored with Roxanne Hiltz thrilled my social imagination. Network Nation was a cult classic. Many users of the early computer conferencing and bulletin board systems were singing odes to it, as it opened our eies (pun intended) to new ways of collaboration and social self-organization.

Unbeknownst to me then, the book united my past and future. Before going to India, I was a radical sociologist by training and social change movement organizer by calling. Reading about computer-mediated collective intelligence, self-determination, their methods in the Network Nation, and the case studies illustrating them, pulled me into an adventure of my life for the next 40+ years, which is still going strong.

That book also inspired me to join and dive into EIES as soon as I read about it, where I hung out and learned from its authors, as well as with Steward Brand and other heroes of mine from the counterculture of the ’60s.

One step followed the other, and in 1985, I found myself in a loft of New York’s Greenwich Village, where with a dozen “co-conspirators,” we founded the Electronic Networking Association of computer conferencing facilitators. We aimed “to promote electronic networking in ways that enrich individuals, enhance organizations, and build global communities.”

Fast-forward to CityVoice, 2023

Since I started playing with ChatGPT last year and the subsequent arrival of the variety of AI bots, I’ve been continually intrigued by the multiple talents of my AI mates’ creativity. When that creativity is at its best is more like their co-creativity with our human team.

To learn about the evolutionary potential of AI, I devised an experiment in human-AI collaboration, which would simultaneously illustrate how far we can go together and address a real-world challenge:

I asked Google’s Bard and Anthropic’s Claude: “Can AI agents supporting different roles in a team combine their efforts?” Below are their different but complementary replies. The first is Bard’s the second is Claude 2’s.

Bard’s use cases seem to be slanted to Google’s corporate customers, while the examples provided by Claude 2 tell more about the various capabilities that AI-to-AI collaboration can exercise. Putting the same question to both was a gentle hiring interview that helped me gauge with whom I should work on this experiment. Claude 2 won, so I addressed my next question to it: Can you provide examples of how AI agents have successfully collaborated as a team? Look at its answer:

All those examples are exciting, but inventorying possibilities for AI-to-AI collaboration is only an intermediate step in my exploration. What I am after is gaining more insight into human-AI hybrid intelligence and co-creativity. That’s what the “Dialogue systems” example points to, where “multiple chatbot agents with different personalities and expertise collaborate to hold conversations with human users. They balance topics and hand off conversation smoothly.”

Now, I want to nudge my AI partner towards more concrete examples, and to get there, my prompt should be more concrete, too. What if I included in the prompt such a specific, collaboration-inducing approach as multi-user prompt engineering and also an application area?

To share with my colleagues some of the fascinating possibilities I see in my mind’s eye and make them credible, I need to provide some stories illustrating them. I could surely come up with some scenarios, but I’m more curious about what Claude 2 has to say. Here goes my prompt: Write a scenario in which a team of human and AI agents is tasked to use multi-agent prompt engineering to support participatory democracy in cities. And here comes its answer within a few seconds:

I highlighted “vector” in “vector database” because that’s a new concept to me, and I sense that understanding how it works will be instrumental to my experiment. But not now; I will have to get back to it later with a more tech-savvy colleague, human or AI. For now, I want to dive deeper into the CityVoice scenario, so I click on one of the questions suggested by Claude 2 itself:

That sounds practical and plausible, but I want to know how a team of action researchers who would work on this project could enlist the AI team to secure the presence of marginalized voices. As it would read my thoughts, my AI partner offered a couple of highly relevant follow-up questions, from which I chose this one:

Those examples certainly provide some helpful, actionable ideas. However, to realize them, we need to build partnering relationships with local groups. I’m wondering whether Claude 2 has something to say about it:

Thanks to this conversation with my AI partners, what started as a thought experiment evolved into an inquiry and a seed idea for an action research project with a mixed human/AI team.

If and when the project attracts enough attention and energy from people who care to cultivate AI for the common good, it can deliver tangible results to the participating city communities. Given our appreciation of the project’s broader potential, it can also become a replicable prototype for cultivating collaborative hybrid intelligence (CHI) and human-AI co-creativity.

From Hangzhou through Nesta into the presence and beyond

The flow of this writing, from the thought experiment to the concept of the action research project, moved surprisingly fast. Possibly because its starting point has never really left me since I introduced an embryonic version of it in a keynote at the International Conference on Innovation and Knowledge Management in the Asia Pacific, in Hangzhou, China, on Nov 1, 2018.

The full title of my speech was Creating a Meshwork of Communities of Practice for Unleashing the Emancipatory Potential of AI-Enhanced Collective Intelligence. One of the issues I presented was “The ABCD of civic technologies — connecting them with Artificial intelligence & Big data in City Democracies.”

The focusing question of the issue at that time was this: How to prototype use cases that combine the power of Linked Open Data and AI with collaborative civic technologies for strengthening real democracy in the political ecosystem of the current shift from a planet of nations to a planet of cities?

Convinced about the power of the role that AI may play in boosting the Collective Intelligence of organizations and communities, I convened an AI4CI consortium with which we submitted a grant proposal to Nesta’s Center of Collective Intelligence Design. We received excellent feedback, but the proposal didn’t go through. In 2018, it could have been a bit ahead of the times, but the research and the generative conversations that went into creating the proposal continued as an underground stream. Now it’s resurfacing again and gathering new tributaries. Will you connect with it?

* Excerpt from the forthcoming book Rise of the Compassionate AI.