On 11 March 2019, Harold Channer invited me to the studios of MNN (Manhattan Neighborhood Network) in New York City to record two one-hour editions of the TV program “Conversations with Harold Hudson Channer”. To expand my thinking on topics from the second interview, here are four addenda:

• The Thinking Society
• Other references from the interview
• A brief Critique of AI (Artifical Intelligence)
• My addenda to interviews with Harold Channer to date

The Thinking Society

In 2010, I started attending and then organizing events at The Greater Philadelphia Thinking Society (which was then called “The Ben Franklin Thinking Society”). The group was originally intended, by its founder Michael Tweed, to be loosely modeled on Ben Franklin’s Junto, a club for social improvement, and college dorm chats about life and learning.

The Thinking Society has become a forum for the exploration of ideas where groups of people learn together on an on-going basis. Events are dynamic: the spontaneous responses of others frequently surprise us and invite us to consider perspectives and considerations that we would not have imagined on our own. Each event provides participants many opportunities to realize new insights and possibilities, and to build new meanings. My experience organizing more than 150 such events has gradually convinced me that human thinking is fundamentally a social affair. Yes, of course, each of us has unique experiences and innate faculties for imagination and discourse that provide us with the ability to develop new ideas and capabilities. But to enrich our thinking, to sharpen it with new perspectives, and to strengthen our ideas with the nuanced considerations of others, we all depend on a community of fellow-explorers.

Although it is worth addressing Harold Channer’s concern for group think, a pathological convergence of thinking that most groups are prone to, at the Thinking Society there are at least three protections from this defect. First, our events are open to the public ensuring a continual variation of participants and perspectives. Second, each event considers a different topic and so there is little opportunity for a too narrow convergence to emerge. Finally, since our events are explorations, they broaden and deepen our thinking without insisting on a conclusion. That is, our events usually end with everyone having different takeaways and different ways of thinking about the subject. There are forces that do tend to narrow the diversity of our participants and ideas can converge in ways that bypass thoughtful exploration, so the Thinking Society like all organizations needs to continually guide its development and to root out any such pathologies.

Gradually, my practice in organizing events at the Thinking Society and in conceptualizing the nature and purpose of the organization has changed. Instead of mainly organizing topics on my interests, I have been exploring subjects I don’t know much about and exploring areas of weakness and blindness in my background. To encourage everyone to think through issues afresh, I prefer significant topics approached from unusual angles. I started reading a wider range of books and genres and taking diverse free on-line courses to further expand the scope of the topics I organize. The process has significantly improved my skills at comprehensivity, Bucky Fuller’s name for being “adequately macro-comprehensive and micro-incisive” (as he put it in Operating Manual for Spaceship Earth).

Recently, I have been working to imagine how the Thinking Society idea can be expanded to a global scale. How can it more effectively redress the crises of social fragmentation and divisiveness, over-specialization, mass misinformation, and the disintegration of our sense of purpose that are endemic around the world? Can group explorations be a powerful kind of social trimtab to help correct these shortcomings? A trimtab is a metaphor for small actions that have big effects, so a social trimtab is a social action with big effects. The project is still in the design phase, but an extensive draft for the initiative, Collaborating for Comprehensivism, invites your participation and feedback. See the section Why it works for an expanded explanation of social trimtabs.

What would interest you most about participating in Thinking Society events?

What do you think about the idea of Collaborating for Comprehensivism? What improvements for that project would you recommend? What else would it need to offer to motivate you to organize events for either the Thinking Society or Collaborating for Comprehensivism?

Other references from the interview

My reference to Bucky’s quote that “the subjective and objective always and only exist” is discussed in depth in my essay on the subject: The Objective, The Subjective, and The Nature of Design Science.

A sample Stephanie Kelton lecture is “But How Will We Pay for It? Making Public Money Work for Us”.

I learned about Keynes’ realization that money and thereby our modern economy is an invention of government from David Graeber’s “Debt: The First 5,000 years”.

Elinor Ostrom and others wrote Going beyond panaceas which summarizes theoretical and empirical studies that collectively show that cure-all solutions tend to fail. Ostrom was a pioneer in developing better practices to govern social-ecological systems. Ostrom was also an important advocate for polycentric governance where multiple agents and multiple levels of hierarchal structure successfully work to govern even fragile social-ecological systems.

The late Kenneth Taylor refers to human beings as “norm-mongering creatures” in this nice short video on Normativity.

A brief Critique of AI (Artifical Intelligence)

Since reading Gödel, Escher, Bach by Douglas Hofstadter in High School and taking a course on Artificial intelligence (AI) in college, I’ve been leisurely following AI. Here is a somewhat expanded version of the critique of AI that I tried to sketch in the interview.

It seems to me that intelligence involves purpose. To my knowledge no one has yet proposed an algorithm that confers general purpose-driven behavior into computers. The kind of AI we have today performs particular human skills like image recognition or driving. Sometimes the computer performs better than humans (chess, long and tedious mathematical calculations, etc), sometimes it performs well enough to relieve humans of some of their drudgery, sometimes it underperforms humans (self-driving cars in heavy rain). But none of the AI we have today has the means to imagine and create purpose out of input datasets. If intelligence is about purpose, then the current scope of intelligence in AI must be seen as much more limited than what we see in human purposes and human intelligence.

We can characterize all purpose-driven behavior as design. In the exquisitely provocative book “The Design Way” by Harold G. Nelson and Erik Stolterman, it is asserted that “Designing is like laying track for a moving train while on board”. Moreover, design is about first imagining the not-yet-existing and then making it real which involves many kinds of judgement. To build actually intelligent AI, we will need to develop algorithms for design that can address all the complexity that the laying track metaphor implies. These AI algorithms will need to imagine the not-yet-existing and to exercise exquisite judgment. I have seen no reports that algorithms are even close to accomplishing any of these things.

In order to achieve AI, we will need to better understand human purposes, intentions, desires, and designs well enough to specify algorithms for these traits of human intelligence. Although that may very well be feasible, I think most contemporary AI is more about hype to stoke consumer and investor interest. It should be critically reviewed as advertising and marketing, not breakthroughs in computer algorithms.

I am aware of the powerful introduction of two new technologies into AI practice starting in the 1990s: the development of large datasets on the Internet and the development of Beyesian inference methods to process those datasets. The combination of the two has produced highly skilled computer behavior in many domains once dominated by humans. This field is called “machine learning”. It has ushered in a profound advance in computer science that is profoundly affecting our lives.

Nevertheless, we should bear firmly in mind the fact that datasets are always historical: we cannot collect data from the future! Therefore, inherently, machine learning is driving by looking out the rearview mirror. Moreover, Cathy O’Neil’s book Weapons of Math Destruction (or her excellent TED talk “The era of blind faith in big data must end”) point out problems with biases in the data itself. Even though those in money-making endeavors based on AI will tend to downplay these problems in their rosy assessments of their products and prospects, we should not lose track of what real human intelligence involves just because their glossy brochures and videos look so impressive.

What do you think are the prospects in the near future for computer algorithms to reach the intelligence implied by a general purpose design capability?

My addenda to interviews with Harold Channer to date

I have written three addenda to my four interviews with Harold Channer. For easy reference, here they are:

• This post.
• Addenda to My First 2019 Conversation with Harold Channer
• Addenda to My Conversation With Harold Channer (from 2014)