Theory Thursday: What Can AI Learn From What You Never Said?

Art by @basilonmypizza: https://lnkd.in/eF8FkWzN - https://basilhefti.ch/

Most of what people think is never published.

Yet AI may increasingly be able to infer the missing part.

We do this all the time and call it reading between the lines. And we use it to solve crossword puzzles.

Facebook also discovered this: people not on Facebook can be inferred from the people around them. Their connections (the network) left enough clues to reconstruct part of the missing person.

Now AI makes something similar possible with opinions: use someone’s public history to build a persona model, then question it to infer what that person would likely think.

Imagine being able to ask:

• What would this CEO think about X?

• What does this customer likely prefer?

• How might this competitor react to our next move?

A recent paper, Talking to Digital Twins: Selective Disclosure and Belief Measurement in Financial Social Media, applies this idea to financial influencers.

The key insight: silence is data.

For most stocks on most days, you hear nothing from a given influencer. In the study, public posts covered only about 15% of these observations.

The researchers used the LLM-based persona to ask the same questions every day, creating an estimate of what the influencer might think even when they had said nothing publicly

The result: the inferred views predicted subsequent stock returns better than the influencers’ public posts did.

For businesses, that opens an interesting possibility: build personas, then actually ask them.

What could you learn about customers or decision-makers from what they never said?

• Art: https://lnkd.in/eF8FkWzN https://basilhefti.ch/