First Impressions: Designing Social Actions for Positive Social Change with Bard

In 2008, after nearly a decade of researching and writing, Liberating Voices: A Pattern Language for Communication Revolution, was published. I was singled out as the author of the book, but the 136 “patterns” it contained were contributed by over 75 people. Each “pattern” in the book (and online) is a seed for generating ideas to help tackle the major issues we face today.

The pattern language concept organizes patterns that are intended to be used together into a coherent whole. It was originally developed by Christopher Alexander and his colleagues to promote the construction of buildings and urban settings that would be beautiful and life-affirming. Since its original publication, the idea has spread to other disciplines including computing. The Liberating Voices book addressed issues related to computing as well, but was mostly focused on working for positive social change using communication.

Fast forward to early 2024 where AI—especially in the media—is rapidly gathering steam. AI, according to its advocates, promises many things. Some of these are fairly reasonable, some too good to be true, and some, on the other hand, are flat-out apocalyptic.

At the suggestion of Aldo de Moor, the co-author of our follow-on pattern language book (still in work), I went to Bard, the deep learner brought out by Google, who provided the deep pockets to fund it. The notion was that Bard’s capabilities could be used in some way or another to help our project, making pattern languages more accessible, effective, and popular. Of course, AI, like computing itself could be useful for addressing nearly everything—including positive things. Unfortunately, as we know, computing applications tend to have a distinct partiality for those promising deep ROIs.

That caveat notwithstanding, I offered up the following prompt for Bard’s consideration: “Select three patterns from the Public Sphere Project Liberating Voices pattern language and incorporate them into a creative and effective community project.” After seemingly no time at all, I received this response: “Here’s a creative and effective community project that incorporates three patterns from the Public Sphere Project Liberating Voices pattern language.”

(In retrospect, maybe I should have said “novel” instead of “creative?” I’m also wondering how it knows those will be effective. Can I get that in writing?)

Bard, usefully, also provided a title, “Community Climate Action Hub” and an objective, “To foster a vibrant and collaborative hub for community members to learn, connect, and take action on climate change.” It also identified these patterns: Community Inquiry, Illegitimate Theater, and Open Source Search Technology. (Funny that it used a closed-source search technology to advocate for an open one!)

(Now I’m wondering whether these were the just first ones it ran, into or did it actually find the other 130 or patterns less useful?)

It went on to say how it would use each of the patterns. It also included seven actions under the “Project Implementation” heading, including specific tasks like “Organize workshops, town hall meetings, and door-to-door campaigns to inform residents about the project and encourage participation,” “Form diverse groups of community members to participate in regular inquiry sessions, facilitated by trained moderator,” and “Collaborate with local artists, theater groups, and community members to create engaging and impactful performances.”

I should mention that Bard created two other drafts in addition to the community climate action hub. Each was equally well-organized and used slightly different approaches, including project title, purpose, list of patterns, project activities, expected outcomes, and others. And one of the drafts used Cyberpower instead of Illegitimate Theater.

Was I impressed? I had to admit that I was.

Did I, Aldo asked, think it was cool? That was not actually my reaction. I was gobsmacked. I did not not think it was cool, but my mind was unsettled. I did not actually know what I thought.

Fairly soon the thoughts did begin to congeal.

Bard’s response was actually better than expected, coherent and viable. The patterns that it selected were all patterns from the book. Not hallucinations at all. And I was happy that it had actually taken the time to consult my book. (But did it like it?! And what about the other 130+ patterns?!)

But why did I hesitate in my response? Was it because I thought it was not cool?

Sadly, my first reaction was that Bard’s response was actually better than many of those that participants have developed over the years in the workshops using the Liberating Voices patterns. (Faster too!)

The response reminded me of the smart kid in the class who apparently has some acquaintance with the material and makes a point of saying something that is actually relevant every class session because they know the professor requires class participation.

(And I can see a more recent version of that smart kid, in a collaborative workshop setting, not actually paying attention or contributing to the group work, but summoning Bard from the sidelines to more easily provide “their” answer.)

As my initial dazzlement gave way to more sober consideration, the first thought that came to mind is that coming up with an “answer” is a very far cry from actually following through with the suggestions—actually doing the work— especially with the diligence, perseverance, and nuance that are required to make progress in the real world on problems thrown up by the real world.

It seemed to me that the freshly minted version, however impressive, ultimately accomplished very little. While it could be good for people who are looking for ideas to consider, it also seems like it’s implicitly suggesting your work is trivial: Why should you bother putting together your own proposal when Bard spits out, without breaking a sweat, proposals that we might not ever come up with—and obviously not within the microseconds that Bard extended before making its.

And, as mentioned above, Bard’s responses are certainly no substitute for action. And the more people who believe that if such answers to all of our issues are available via a few keystrokes, the less likely we are to address our problems collectively. We would simply be awaiting commands. (A corollary of that perspective is the even more unlikely idea that people (such as politicians) would do the right thing and adopt the right “answer.”)

If things would go the way I’d like, people would actually go through the work to come up with the ideas—or, ideally, something better—or ones that they were more excited about or ones that would better fit their particular needs or environment better. I would prefer that people would take the time and effort to plan these things out by themselves. That, of course, may be a bridge too far. Why would anybody want to spend more time to actually come up with something that seems like less? My response might be that the individual and group skills and creativity were exercised (and cultivated and honed) and that in itself was a good thing. Then their response to that might be, should people also do multiplication by hand or in their head or should they use a calculator, which is faster and more accurate? But while multiplication is purely mechanical, I would still argue that people still need to have at least internalized the idea of multiplication, having some notion of what it is and how it works, and when and where it might be useful.

No, I would not banish calculators, but coming up with ideas and building group capacities are not purely mechanical. They are developed through use and experience, and having a proposal dropped from the sky (or cloud) will not help with that—and may, in fact, hinder it. (No more than a set of links is an adequate response to a question for a critical essay— which has not always prevented students from trying it.)

And, of course, all of the words assembled in Bard’s responses could be found in the various patterns. This should not be a surprise, the patterns are fairly similar in their broad perspective as they generally rely on people to assume some responsibility, have some desire to play positive roles, and to work together intelligently with others. Having said that, Bard paraphrased (and plagiarized?) more quickly that any human could possibly do. And, for better or worse, Bard apparently drew on verbiage that was fairly local to the patterns that it found. What does that mean? For one thing, a human after being exposed to the Liberating Voices patterns might pass them to others. Bard, however, which presumably merely throws our own text back at us, would not deign to suggest the patterns to other Bard users, no matter how much I think it is a good idea. (Next version?)

While the approach I took, asking for some creative uses of the patterns, in the right hands, could be used to augment ideas that were already on the table or to suggest ideas that people could build on, I suspect that a more common use is for people to believe they now have answers, when “answers” are not really what are needed. The problems we face do not have “answers” in the sense that they can be “solved” by some mechanical approach (whether produced by machine or humans). The essential resource in any effort to address the issues of the day is the ability for people to work together. To the extent that Bard and the like encourage that determines the value that I would ultimately ascribe to it.

Finally, as you may have noticed, these issues under consideration are predicated on a preference for human reasoning over machine reasoning. But is this faith in humanity warranted? Having read the newspaper and leafed through history books, I would not say that I have faith in people to do the right thing. I would, however, state that I do have faith in the potential of people to do the right thing. Often, however, even without apparent opposition, our untold counterproductive habits of mind and behavior stand in the way of this potential. Nevertheless, surrendering human autonomy is a terrible idea and it is not easily reclaimed if lost. Neither AI, nor any technological approach, is able to bypass or override these hard truths.

Douglas Schuler is professor emeritus at the Evergreen State College. He is author of New Community Networks: Wired for Change (Addison-Wesley), Liberating Voices: A Pattern Language for Communication Revolution (MIT Press), and many articles that explore the relationship of computing, society, and the environment.