Newshot

Being right is not a strategy: Make them feel something

Summary

This is a cleaned-up version of a talk I gave on communication to technical researchers at Constellation in Berkeley. The event was organized by Alexandra Narin and Nicole Xing . I studied physics, and I thought like a physicist. I then became a journalist, and I did journalism like a physicist would do it. I sought the truth, and put it out there. In Colombia I uncovered a paramilitary massacre and tied it to a multi-billion dollar company. In northern Iraq, I trampled through an Islamic State munitions dump to work out the provenance of their weapons. And, in my most black and white investigation, I got hold of the Venezuelan government's own spreadsheet of results for a national election. Its numbers weren't the numbers they'd announced. I didn't post my stories on LessWrong, or my blog, but on the front pages of the biggest news outlets in the world. But nothing changed. So what did I get wrong? I'd done the reporting, which is my version of the science we do here. And I'd publishe…

Original text(via LessWrong)

This is a cleaned-up version of a talk I gave on communication to technical researchers at Constellation in Berkeley. The event was organized by Alexandra Narin and Nicole Xing.

I studied physics, and I thought like a physicist. I then became a journalist, and I did journalism like a physicist would do it. I sought the truth, and put it out there.

In Colombia I uncovered a paramilitary massacre and tied it to a multi-billion dollar company. In northern Iraq, I trampled through an Islamic State munitions dump to work out the provenance of their weapons. And, in my most black and white investigation, I got hold of the Venezuelan government's own spreadsheet of results for a national election. Its numbers weren't the numbers they'd announced.

I didn't post my stories on LessWrong, or my blog, but on the front pages of the biggest news outlets in the world.

But nothing changed.

So what did I get wrong? I'd done the reporting, which is my version of the science we do here. And I'd published. And when things did finally shift in those countries, it had nothing to do with my work. They shifted due to political whims or military defeats.

What I realized was that facts alone rarely change anything. Does that mean facts don't matter? No, but they're not enough.

When you're communicating with people, you need to meet them where they are, and you need to make them feel something. People don't remember what you say, but they remember how you made them feel. That line—not from Maya Angelou, but rather Carl W. Buehner—is what we must remember.

Sydney just gave a wonderful, rousing speech about how we should communicate. I can't quite remember what she said, but I want to go out and do something.

Jacob Coxon's resignation did it too, not because of the technical detail, but because he told a story. He thought his work might be hurting humanity. So he faced a conflict: Stay, take the money, keep the boss happy? Or walk out and say so? Everyone has faced a small version of that. You could feel it.

Buck's post did it. Buck is a numbers guy, and he gave a number: one in three for superintelligence within a year. He also said he was sad. When I interviewed him afterwards, he said he was scared. Yes, people will quote the number, but they will do so because they felt his fear.

You should meet people where they are. If they care about their job, don't start by telling them they're wrong. Validate that fear, and hold their hand to take it to what you want to discuss.

And bring in some passion! I sit in talks detailing scenarios for humanity ending by various mechanisms within a few years. It's all abstract. What actually happens? What happens to me? Make it concrete. I've no idea, so I feel nothing.

Then comes the last slide. "Thank you to my twenty collaborators, and to an institute in Switzerland for five hundred dollars." Excuse me, the world is ending and we're doing acknowledgements?! If you tell me humanity is ending in the voice you'd use for a job interview, I can't believe you. Nobody does. You think you sound professional, but you sound inauthentic.

So tell stories. What are you seeing in your research, day to day? Why does it scare you? How did you, personally, end up caring about this stuff?

I didn't come into this field caring about AI safety. I wanted to sit in a corner and work out how these models work inside through interpretability research. I really wanted to do research! But, the people in this room talked me into understanding what was happening and worrying. I began to care about AI safety. That's my story. It's true, and, in persuading people, it beats any graph I've got.

Say what you believe. Say that it frightens you. Say what you want done. The caveats do matter, but put them in the appendix, or in the paper. They're not for this conversation or post.

I spent a decade giving the world facts, and it changed nothing. Let's not make the same mistake.

The text above was extracted automatically from the source page and may miss images, tables or code blocks. The original is authoritative.

Published by LessWrong
This site only summarises and aggregates; copyright stays with the original author. Headlines and summaries follow the source — where they differ, the original wins.
Read the original ↗