Searching for searching

I parked my car this morning in the village and walked down to my favourite coffee place for an espresso. Every one of the three conversations I overheard was about people discussing the pros and cons of ChatGPT. Pros seem to be that “it helped me to know what to ask for when I talked to my car insurance company” and cons are mostly “how do we know that any of this is real?” More seriously I’m sitting near folks who work in the arts and the looks on their faces are of the deepest concern. They use it. For ideas, for a writing prompt, but the times they have used it to write dialogue, they can spot how crappy it is. At the moment.

My earliest post about was Google was from 2002 when it was an insanely useful tool for searching the web. “Google cooking” was a simple game where one entered in a list of ingredients and it returned a list recipes. It was novel at the time. Great for weeknight dinners. Another game was called “Googlewhack” whereby one would try to construct a two word search term that resulted in only one result. You can’t play that one anymore.

The complete enshittification of search engines, combined with web content that has been generated by robots in order to sell stuff is increasing turning web-search an absolutely useless activity. I just use my search engine (DuckDuckGo) as a collection of bookmarks now. It is hard to do any meaningful research anymore, and so we turn to ChatGPT for answers. And ChatGPT is out there learning the questions we ask. Something sits weird with me when I think about how while Google learned the answers we like, and AI is learning the questions we ask.

The questions are important, as is the way we ask them and to whom we ask them. Sonja today writes about the questions that help us discern a direction, which is different from finding a way. Sometimes we don’t even know what the direction is although we can discern that wherever we are right now, somewhere else is better. Thinking about that and talking about it together is an essential human capacity and it’s a pretty fundamental part of how we work with teams facing complexity. There is an art to asking to right kinds of questions and thinking about them together that reveals a deeper level at which affordances and opportunities might exist. Sometimes getting unstuck means drilling down and not reaching out.

Collaborative outcomes are emergent properties of discrete human systems of encounter and meaning-making – “dialogic containers” I call them. If you are a leader seeking a course of action, you might get some good ideas by submitting notes and documents and harvests into a large language model to suggest possibilities. In fact, you could even have your team members do that on their own and bring the output to a meeting to talk about what they have found. My hypothesis is if you continue to do that without involving humans you will end up with an endless set of ideas and possibilities, but you will miss the co-creationi and co-ownership that makes sustained effort possible in a particular direction. I can’t yet see how large language models can surface a consensus that will inspire collaborative action. Deep meaning and commitment to one another is produced by the people within the container who discover something between them that is worth trying, worth pursuing together. Calls to action are far less sustainable than co-creation of a direction. Even if, and perhaps especially if, such a direction is deeply flawed to begin with. There is nothing better than failing together and then finding a way forward to build cohesion.

I might be wrong in the future but in this moment, systems-complexity and anthro-complexity are different and humans experience emergence differently from mechanical systems, even those that are capable of learning. Dialogue practitioners base their practice on this idea; that no matter how great the ideas are, nothing gets sustained in human systems without the intangibles of co-ownership, meaningful engagement, and dare I say, at some level, love.

Although, who knows.