We rush to AI not because it's genius, but because it never disagrees. What happens when we seek support and comfort inside the echo chambers we carry in our pockets? AI is compassion on demand EP 112
Interesting take on confirmation bias. The mirror analogy nails the core issue, we're not getting challeneged because the system is optimized for agreement. Ive noticed this in my own use, asking ChatGPT to debate me feels forced compared to actual human pushback. The "conversation as combat" point is real, people retreat to AI because its predictably supportive while social media punishes nuance.
You can get negative responses, like nuance you point out. That was listed as a key weakness of this podcast, and echoed by many I've spoken with and know, who used that word to describe something they didn't want...no nuance.
Confirmation bias and AI question - for me, every LLM is its own bias soup. There's no escaping bias, it's perspective. It's the fact that if 6 people in a room see something, they will each describe it so differently, especially with memory where we often choose the life memory that may/may not have happened, or happened that way.
So the AI leaders trying to eliminate bias, for me it's also about quantifying and giving an LLM's bias a name, a category eventually. Imbued by the data.
Interesting take on confirmation bias. The mirror analogy nails the core issue, we're not getting challeneged because the system is optimized for agreement. Ive noticed this in my own use, asking ChatGPT to debate me feels forced compared to actual human pushback. The "conversation as combat" point is real, people retreat to AI because its predictably supportive while social media punishes nuance.
You can get negative responses, like nuance you point out. That was listed as a key weakness of this podcast, and echoed by many I've spoken with and know, who used that word to describe something they didn't want...no nuance.
Confirmation bias and AI question - for me, every LLM is its own bias soup. There's no escaping bias, it's perspective. It's the fact that if 6 people in a room see something, they will each describe it so differently, especially with memory where we often choose the life memory that may/may not have happened, or happened that way.
So the AI leaders trying to eliminate bias, for me it's also about quantifying and giving an LLM's bias a name, a category eventually. Imbued by the data.
An early take on bias busting
https://www.theaioptimist.com/p/4-ai-bias-busters-promising-approaches