AI in Conferences

Overview

Navigating AI-generated content in talk submissions and event management.

How conferences are using AI

• Documentation - AI helps write docs that people avoid writing manually
• Graphics and design - Works okay for smaller events. Bigger events prefer human design.
• Spam detection - Some use AI to flag suspicious submissions
• Abstract editing - Some consider it, but people have concerns

AI used by participants:

• Self-hosted (open model) or Claude/Copilot etc.
• Should you use open models for OSS projects, or can you use paid ones?

AI-generated submissions

This is happening now. Some talk submissions are entirely AI-generated.

Detection and policies:

• Kubernetes requires attribution, bans AI-generated code comments
• Grafana tries to detect and flag it
• Hard to detect consistently
• Even if you detect it, is that bad if the content is good?
• Some people submit more when AI makes drafting easy (increases spam)

Ethical questions

• Is it okay to rewrite someone’s abstract with AI? Feels weird.
• Should speakers disclose AI use?
• Fair if some speakers get AI help and others don’t?
• Non-English speakers might benefit from AI translation, but most AI is English-trained

Practical approaches

• Interview speakers before accepting talks. See if they actually know their stuff.
• Speaker coaching - Helps with abstracts and confidence
• Clear policies - Just say what’s okay and what’s not

What projects are doing

Kubernetes:

• Clear guidelines requiring AI attribution
• Bans AI-generated code comments

General approach:

• Rather than strict bans, ask reviewers to check content quality
• Flag low-effort submissions regardless of tool used
• Be transparent, don’t punish disclosure
• Focus on community value, not purity of origin

The reality

This is still forming. No one knows the right approach yet.

Blanket bans probably aren’t practical. Being transparent helps. Different events will have different rules based on their community values.