Basement Whatever

We have a competitive Illinois primary coming up in March, so I wanted to see if generative AI could help quickly get to a rough estimate of the best candidates to vote for. It certainly can – especially when you use it to filter out the most insane candidates with a goal of keeping them away from power.

Rather than getting overly complex, I decided I would try a simplified process where I filtered out the most insane candidates. I picked a position that is insane, and started from there:

1. Go to perplexity.com
2. In the [PARTY] primary on March 17 for Illinois [POSITION], can you tell me who is the most [POSITION] and who is the least [POSITION]?
3. Perplexity won’t be able to rank this precisely, but it will get close. And it will do this by looking at who gets endorsements from which groups that are most aligned to the position you find repugnant. If you want to go deeper, you can further filter by exploring candidates that have affinity with those terrible groups. But for my purposes, it was enough to take the least insane on any given position.

Using this prompt allowed me to quickly pull together a ballot of candidates that are the most sane according to my preferences. Note that I went position by position, rather than trying to do them all at once, as I believe from my experiments with Replit that this is the best way to get the most successful outcome.

Sidenote, during this research I came across the site ballotready.com, which is really impressive in terms of the research it provides on candidates. I’m going to add that site to my ongoing list of net-positive technology advancements. Also, you don’t need to provide an email address. It will work without one.