Why LLMs Trust What Others Say About You Online (Full Transcript)

LLMs form brand opinions from web-wide sources; third-party mentions often outweigh self-written bios, shaping knowledge panels and perceived credibility.
Download Transcript (DOCX)
Speakers
add Add new speaker

[00:00:00] Speaker 1: This is a visual example of what the LLMs are looking for. So this is what the LLMs are pulling from and what they think of you. This knowledge panel is showing you, like, hey, this is what we think of you. This is what we found from all over, all of our sources on the internet. So you'll see there's like my Twitter feed, my, you know, my appearances on Lawyerist. And it's usually what it wants to see is what everybody else is saying about you. It finds that to be more true. And so these LLMs also will find what everyone else is saying about you to be more true than this garbage that you may be churning about yourself on your own about page.

ai AI Insights
Arow Summary
The speaker explains that large language models (LLMs) build an understanding of a person or brand by aggregating information from across the internet. A “knowledge panel” reflects what the model believes based on external sources (e.g., social media, media appearances). The key point is that third‑party mentions and what others say about you are treated as more credible than self-written claims on your own website, such as an “about” page.
Arow Title
How LLMs Form Opinions Using External Signals
Arow Keywords
LLMs Remove
knowledge panel Remove
online reputation Remove
entity understanding Remove
third-party sources Remove
social proof Remove
Twitter Remove
media appearances Remove
about page Remove
credibility signals Remove
Arow Key Takeaways
  • LLMs aggregate information from many online sources to form a profile of you or your brand.
  • A knowledge panel represents what the model has inferred from the wider web.
  • Third-party mentions (press, interviews, social profiles) are weighted as more credible than self-authored bios.
  • Improving how others describe you online can influence how models represent you.
Arow Sentiments
Neutral: The tone is explanatory and instructional, with a mildly critical edge toward self-promotional website copy (“garbage”), emphasizing credibility of third‑party sources.
Arow Enter your query
{{ secondsToHumanTime(time) }}
Back
Forward
{{ Math.round(speed * 100) / 100 }}x
{{ secondsToHumanTime(duration) }}
close
New speaker
Add speaker
close
Edit speaker
Save changes
close
Share Transcript