meta_pixel
Tapesearch Logo
Log in
Question Everything

Can AI tell us what stories to look out for? A live experiment.

Question Everything

Brian Reed

News, News Commentary, Society & Culture, Documentary, Technology

4.6707 Ratings

🗓️ 5 December 2024

⏱️ 29 minutes

🧾️ Download transcript

Summary

Our team at Question Everything has been playing around with a new technology that sucks up tons of social media posts, and then uses AI to figure out what ideas are forming in the shadows of the internet before they hit the mainstream.

Brian interviews a journalist who uses this tech, to see what conversations are brewing right now that we might want to keep an eye out for in the coming weeks.

“Question Everything” is a production of KCRW and Placement Theory.

 

Transcript

Click on a timestamp to play from that location

0:00.0

In the last couple months, our team here at Question Everything has been playing around with this new technology.

0:05.8

It's called Pira. It's a startup. And it uses AI to determine what people are talking about in all sorts of corners of the internet before it breaks into more mainstream conversation.

0:16.2

And eventually the news and real life.

0:18.8

Is there a topic that I can kind of center the demo around?

0:22.1

It's something that's interesting.

0:23.9

This is a training session, Pira's CEO, Welton Chang, gave our staff a couple months ago.

0:29.0

Wilton used to be an intelligence officer for the Army.

0:31.0

Then after 2016, when it became clear that Russia had been meddling in online spaces

0:36.2

to influence American voters.

0:38.6

He left, and he and a colleague ended up joining an NGO to start developing this technology to track security threats online, hate speech, disinformation, threats of political violence.

0:49.5

They started by focusing on what they called unmoderated social media platforms, places like Telegram, 4chan,

0:55.5

parlor that do less flagging and fact-checking of content than Facebook, for instance, or in some

1:00.0

cases, none, because they knew this is where certain ideas and obsessions and conspiracies often start

1:06.3

to percolate before sometimes jumping over to more mainstream platforms like Twitter and Facebook, and then into the wider conversation.

1:13.8

They were hoping this could help them detect threats early.

1:17.4

The tech basically vacuums up tons of posts from across these platforms, a massive amount of data each day, and then they use AI to analyze it.

1:25.6

We take every post and run it through our language modeling capability.

1:29.6

So what that does is it identifies highly negative content, highly offensive, highly hateful, and highly violent.

1:36.9

And it was after Election Day in November 2020 when Donald Trump was claiming he'd won and and that Democrats had committed widespread voter fraud,

1:45.5

that Welton and his colleagues saw the potential power of what they were building.

1:49.6

So the election results were still being tabulated.

1:55.4

And, you know, there was talk of, you know, voter fraud and all this stuff in the lead-up to the actual ballots being cast.

...

Please login to see the full transcript.

Disclaimer: The podcast and artwork embedded on this page are from Brian Reed, and are the property of its owner and not affiliated with or endorsed by Tapesearch.

Generated transcripts are the property of Brian Reed and are distributed freely under the Fair Use doctrine. Transcripts generated by Tapesearch are not guaranteed to be accurate.

Copyright © Tapesearch 2026.