Perry Walker writes:

For this workshop in Hereford we decided to adapt Pol.is, which is a website that:

  • Gets people to make statements about some issue
  • Invites others to say whether they agree or disagree with those statements
  • Uses that information to sort people into attitude groups
  • Then invites people to suggest statements that might command agreement across those groups

We decided to adapt this to a small-scale meeting. We felt that the lack of numbers was compensated for by the opportunity for deliberation.

How did we adapt it? First, we used AI to generate the attitude groups. Lukas Salecker of deliberAIde (www.deliberaide.com) helped us out. He fed 40+ papers on assisted dying into ChatGPT and asked it to identify three equidistant attitudes on the topic. The three were:

  • A. Pro: on the basis of autonomy and compassion for suffering
  • B. Con: on the basis of the risk of a slippery slope and lack of protection for those who are vulnerable, for instance because of disability
  • C. Pro: provided there are strict medical and legal frameworks

The AI also generated an archetype for each position, together with the arguments they would use. For instance, C. was exemplified by Dr Laura Bennett, a palliative care physician. She argued that, “Clear guidelines and robust safeguards can prevent abuse while allowing those who are suffering intolerably to have a choice in their end-of-life care.”

The session, which lasted a bit under two hours, was divided into two halves, separated by a break. In the first half:

  • People used the characters and their arguments to decide where they personally stood in relation to the three positions
  • We made a triangle on the floor, with each point of the triangle representing one of the three positions. People placed themselves on the triangle to show their point of view. We had ten participants in this experiment, and at this point two agreed to adopt a position not their own for the evening, in order to have roughly even groups.
  • People formed three attitude groups to deepen their understanding of their position. They were invited to use the question, “What really matters to me about this issue?” They had additional material to read in the form of cards and newspaper articles.

In the second half, after the break, everyone formed new groups, mixed this time, each containing at least one person from each attitude group. Now we were really into Pol.is territory. They were invited to craft statements that they thought all three attitude groups could agree with.

The 12 statements (four per group) were laid out on tables and voted on, using the criterion: “Can I live with this statement?” People voted with coloured dots, to show which attitude group they belonged to. We then calculated what percentage of each group had found the statement acceptable. The lowest percentage represented the score for that statement. Thus if everyone in two groups voted yes to a statement (100% each) but nobody in the third group did (0%), that statement scored zero, because there was no consensus across the three groups.

Here are the three statements with the highest scores:



Two Column Table

Statement Percentage score of the lowest scoring group
We need to make sure people are heard and views are allowed to be explored with professionals and family 100
An individual should have autonomy over their life, at the critical point in that life. This should be balanced to allow for objective guidance and information on the impacts and outcomes on others, as well as the individual 66
Assisted dying would only be applicable if it was the express wish of an individual who had not been subjected to undue duress/interference 66




We also calculated the scores for the three mixed teams. These were very close: 166, 157 and 149.

As people left, we asked them to give us a little feedback. Of the nine people that did so, the ratings for ‘I learned new information’ were mixed: this was a knowledgeable audience. But on ‘I clarified my thinking’, seven gave the event the top rating, a 5, and two gave it a 4. And for ‘I enjoyed the event’, everyone was a 5. That’s what we like to see!