Last time, we saw a demonstration of AI misinformation and deception, but this is worse. Shadow banning has long been suspected, but it’s hard to prove. Is that nobody malcontent really being shadowbanned, or does he deserve to be on page four of a Google search for his name? This might be another instance of the AI silver lining effect. LMs seem to have no problem spotting these shenanigans.
Key Points:
1. The Personal Touch of Shadow Banning
"Hey, Gemini, about six months ago I was not being shadow banned by Google slash Gemini, and even though I'm certainly not a big deal or a high profile person, I was able to get a reasonable bio about myself from Gemini..."
Our host, Alex Tsakiris, found himself in the crosshairs of shadow banning.
2. The Demonstration
"You said you didn't know anything about this person. And then, when I pasted in the bio, you verified every point of the biography, and even added some new ones..."
Gemini it's pretty loose with the “truthful and transparent.”
3. More Hidden
"This hidden alignment problem is even worse than the misinformation and deception we saw last time… this is harder to spot"
The implications of Shadow Banning within AI dialogues not only pushes the boundary of AI ethical issues, but may present legal problems for those engaged in the practice.
Shadow Banning and AI: When Transparency Goes Dark |03|