Automating Summation — On AI and Holding Responsibility in Relationships
Dullness and Necessity
For years, there has been an argument that robots are well suited to the kind of work that is called DDD — Dull, Dangerous, and Dirty. There is a longer history of that term that warrants interrogation, but we’ll park that for now. In recent years, a fourth D has been added — Dear (or expensive) — to the kinds of work that might be done with automation. While this framing is not a perfect map for what might be automated with AI, it is the category of “Dull” that I’m pulling out for specific consideration.
There is dull work in almost every occupation, and in many civic cases as well. It can take many shapes and forms. A lot of it, but not all, is administrative As someone with decades of experience in both report writing and note-taking, I know and respect much of the art of both. There is a lot of power in note-taking, and also in summarizing what happens in meetings of so many types. Both skills — reporting and note-taking — are important facets in a lot of the ways we do our work. There is also a special skill that is tough to hone without a lot of practice, and this would be the art of summation and/or synthesis.
One of the types of work I was lucky enough to do for several years was facilitating public meetings. The style of running a public meeting that I was taught as method usually involved people sitting around small tables, spending some of the meeting time in small groups talking to each other, and some of the time as a whole room, together. Third Party Public (formerly Swerhun Inc.) has decades of experience doing this, and was where I learned it.
In that process, we had an agenda that we would include in our invites that showed how the time together at a public meeting would be spent. xx minutes on introduction, xx minutes for Q&A, xx minutes to talk about the issues at small tables, xx amount of time to report back from the tables to the room, xx time to wrap-up and talk next steps, etc.
Using this method, we would have a worksheet that helped frame the conversations using discussion questions about the topic at hand (new park, library hours, transportation planning, etc.). We’d have the worksheets available at the door and on the tables so that every single person could write down some of what they thought about the matters at hand and leave it with us as their advice to the government or government agency that was convening the conversation. By being the people running these meetings, we became responsible and in relationship with everyone that participated. It was our responsibility — in that relationship — to take the participant’s time and energy, their knowledge and beliefs, their advice and ideas, and their delivery of these things, and summarize them in a report.
This written version of the worksheet made sure that we didn’t only factor in what we heard said/spoken that night, amongst other things. What gets said, and how it gets said, carries a lot of weight in a room. What is unsaid does too. Then there is the challenge of incorporating what is said only in writing, and how to accurately represent a range of advice from a community.
After a meeting, one of things we would do is collect up all the individual worksheets and get to work on summarizing them. Imagine anywhere from 80 to 200 sheets with handwritten responses on them. Part of the work, and best started the very next day, was to read each one of these sheets and try to organize what was shared (this part of the park needs kid stuff, dogs have a nearby play park, etc.). Some of it had to be organized by topic, some of it had to be organized by frequency (lots of people said y, a few said x, one or two said z).
The one point I want to make here was about the impact of spending time with those worksheets. From the physical act of transcribing the handwriting into a computer to the literal time spent reading and categorizing and indexing — that time spent made those sheets stand in for the participant in a way that I felt because I had to keep spending time with those sheets. Did I love that work? Sometimes, not always. I’m a good typer. Sometimes the help from a colleague in deciphering handwriting led to euphoria on jointly figuring it out. Sometimes the work was just head down, type it up, figure it out, and sort it. Was it dull? Not exactly, but I can see how that description might be assigned to some of it. Was the transcribing process efficient? Not really, there is probably writing to text or arguments to make for handing out ipads, or whatever else. But that inefficiency forced me to spend time with what I had to pay attention to. It was a written record of import, and I had now become responsible to the participants to figure out how to include it both accurately and appropriately in the reporting process.
If this process was done differently, one of the things that would get lost is the visceral impact it has on us to physically spend time with whatever representation we have of each other. In this case, the worksheet, or the notes we took at the meeting — these became and were important to the relationship that I wanted to hold in a way where my duty to be careful and thoughtful was upheld.
That’s the impact on me on the inputs process. Then comes the summation, the work and skill required to try to figure out how to properly summarize what happened in a room, what happened on the worksheets. When I see AI being suggested as a summarizing agent, I’m not only concerned about the accuracy of what is created through the use of automation, but moreso the absence or loss of what does not get done — what is inefficient and what is dull. I’m concerned because in the time-pressured world we live in, where efficiency is a constant measure of our professional capacity, there is every incentive to rid ourselves of this type of work if and where we can.
Then comes the difference between what a machine can summarize of words and what it cannot even consider because of how something was said. We’ve all been in a meeting where the delivery of a remark was much more important than the words used to say it. That is something that I’ve learned how to translate into my reporting. I’m pretty confident this skill is hard to pull off in any automated way.
How to protect necessary inefficiencies and skills in a world where we’re incentivized to get out of the weeds of details, away from anything dull? I’m not sure. And I know, with certainty, that this is a perfect example where regulation and law won’t catch the issue, unless perhaps in a very small number of cases.
The importance of how we manage information as part of duties and relationships is high.
