Last week, I promised an update about the progress Seattle Worldcon 2025 has made regarding our next steps related to remedying our mistakes related to the use of ChatGPT in panelist vetting. Much of this update can be summed up as “we’re waiting to hear back from the people we have invited to help.”
- Regarding re-vetting, we have invited two people, new to our team, to join, and we are waiting to hear back from them. We are still searching for at least three to four more people to join that team. If you would like to volunteer, please email feedback@seattlein2025.org. This new team will be working with our existing program team but be reporting to the chair.
- We have reached out to a team of two people with prior Worldcon programming experience to audit our program process and the remedial steps. We are still waiting to hear back from them.
- We have processed all refund requests that we have received; former members will be receiving them this week.
Our next update will be once we have identified who is helping with re-vetting and performing our program audit or in three weeks, whichever comes sooner.
I hope this blows over for you. I might have recommended your original post amount to something like this:
“Hey, we found out some of our volunteer team used ChatGPT to help them with the incredibly daunting task of finding any hidden potential code of conduct violators among potential panelists. We didn’t ask them to use that tool, but we don’t micromanage so they did. We’ve told them that many in the membership community oppose the use of those tools, so they have stopped. We’ve evaluated what took place to assure no mistakes were made due to LLM hallucinations, and right now it all looks good, but we will double check. See the technical details under a different post. Sorry about all this, but convention management is an all-volunteer task with lots of cat herding and mistakes will happen, we hope you’ll be forgiving.”
Though some people would not have been forgiving.