Trending In Ed with Mike Palmer Revolutionizing Research with AI | A Conversation with Subconscious AI Founder Avi Yashchin
Apr 3, 2025
In a compelling discussion, Avi Yashchin, a serial entrepreneur and founder of Subconscious AI, shares his journey from high-frequency trading to groundbreaking AI research. He highlights the use of language models to simulate social-science experiments, exploring both their potential benefits and ethical complexities. Key topics include the critical role of human oversight in AI education, the challenges of proving social media harms, and the need for transparent AI practices. Yashchin also emphasizes the importance of directing AI towards pro-social initiatives like climate mobilization.
AI Snips
Chapters
Books
Transcript
Episode notes
LLMs As Digital Twins For Research
- Subconscious AI uses LLMs to simulate human subjects and run randomized controlled experiments at scale.
- They found LLM responses were near indistinguishable from human responses in many psychology and economics studies.
Turn Model Bias Into Measurable Data
- Model confabulation mirrors human confabulation and bias, so biases can be measured rather than dismissed.
- Subconscious AI surveys multiple LLMs and aggregates results to quantify model differences.
Build Boldly, Prioritize User Benefit
- Don’t be shy: use accessible LLM tools to build products and create value quickly.
- Prioritize doing the right thing for users and avoid solutions that harm learning or reward cheating.
