Learning Module: What the Experts Actually Think (And Why They Disagree)

Here's a secret: asking AI experts "When will AGI arrive?" is like asking medieval cartographers about the edge of the world. Everyone has an opinion. Nobody really knows. But HOW we ask matters more than you think.

The Art of Asking the Impossible

Imagine trying to survey experts about something that doesn't exist yet. That's the challenge researchers face when gathering AGI predictions. It's not like asking doctors about a new treatment or economists about next quarter's GDP.

We're asking computer scientists to predict when their creations will outsmart them. Talk about awkward.

Who Counts as an Expert? (Spoiler: It's Complicated)

First challenge: Who gets a vote? The field is wonderfully messy—a mix of:

• Publication powerhouses: The professors with papers in Nature and NeurIPS
• Citation celebrities: Researchers whose work everyone references
• Industry insiders: The folks actually building these systems at OpenAI, DeepMind, and Anthropic
• Global voices: Experts from Beijing to Berkeley (geography matters!)
• Fresh perspectives: Junior researchers who grew up with transformer models
• Cross-disciplinary minds: Neuroscientists, philosophers, even psychologists
• Method masters: Theory builders AND hands-on engineers

Here's the rub: Miss any group, and your survey is like a jury without key witnesses. Include everyone, and you might dilute expertise with enthusiasm.

The Questions That Make or Break Your Survey

Remember the dress that broke the internet? Was it blue/black or white/gold? AGI surveys face the same perception problem, but with trillion-dollar consequences.

Every word matters when you're asking about the future:

• Say "AGI" vs. "human-level AI": Get different answers by 10-20 years
• Ask for percentages vs. dates: Watch confidence levels plummet
• Mention specific capabilities: Suddenly everyone's timeline shifts
• Frame it as opportunity vs. risk: Optimists and pessimists emerge
• Start with recent AI breakthroughs: Hello, anchoring bias!
• Ignore cultural context: Miss how different cultures view intelligence itself

The best surveys? They're like good therapists—they know how to ask questions that reveal what people really think, not what sounds smart at conferences.