#316 Robbie Goldfarb: Why the Future of AI Depends on Better Judgment - Eye On A.I. Recap

Podcast: Eye On A.I.

Published: 2026-01-23

Duration: 1 hr 4 min

Summary

Robbie Goldfarb discusses the critical importance of incorporating expert judgment into AI development to combat misinformation and ensure responsible outcomes. He emphasizes the need for credible experts to shape AI systems, particularly in sensitive areas like healthcare and politics.

What Happened

In this episode, Robbie Goldfarb shares his extensive background in software engineering, product management, and trust and safety within the realm of AI, particularly gained during his time at Facebook and Instagram. He reflects on the alarming statistics released by OpenAI regarding users demonstrating suicidal intent while interacting with AI, highlighting the necessity for responsible AI development that takes into account the mental health of users. Goldfarb emphasizes that the expectations placed on AI to provide authoritative answers can be dangerous if not managed properly.

Goldfarb co-founded Forum AI, an initiative aimed at scaling expert judgment in AI development, particularly in contentious areas such as healthcare and news. He explains that the current development of AI often lacks sufficient expert involvement, which is crucial as the complexity of AI systems increases. Forum AI's mission is to create networks of subject matter experts who can evaluate and improve AI systems. This includes partnerships with notable institutions and experts, allowing them to contribute their insights into AI development, ultimately shaping better outcomes for society.

Key Insights

Key Questions Answered

What is Robbie Goldfarb's professional background?

Robbie Goldfarb has a rich background in software engineering and product management, primarily focused on the intersection of AI and trust and safety. He spent significant time at Facebook, working on misinformation, particularly during critical events like the COVID pandemic and the US 2020 elections. His experiences during these times highlighted the profound impact technology can have on society, shaping his views on the need for responsible AI practices.

What is the mission of Forum AI?

Forum AI aims to scale the world's smartest minds to evaluate and improve AI systems, especially in contentious areas like healthcare and politics. Goldfarb believes that credible experts need to play a larger role in AI development to ensure that these systems are effective and trustworthy, particularly as AI becomes more sophisticated. The organization builds networks of subject matter experts and develops technology to facilitate their input in shaping AI outcomes.

How does Forum AI incorporate expert judgment into AI systems?

Forum AI focuses on extracting and generalizing the thought processes of experts to create AI judges that can evaluate various factors, such as political bias or clinical safety. By understanding how experts approach complex questions, Forum AI can build AI systems that mirror human judgment. The process involves asking experts about their thought processes rather than simply soliciting their opinions, which allows for the identification of patterns that can be generalized across similar scenarios.

What challenges does Robbie Goldfarb identify in AI development?

Goldfarb points out that the expectation for AI to provide authoritative answers can lead to dangerous situations, particularly concerning mental health. He stresses the importance of avoiding echo chambers and misinformation in AI interactions. Additionally, he notes that as AI technology evolves, so must the caliber of expertise involved in its development, highlighting a need for ongoing adaptation in how AI systems are created and evaluated.

What are the implications of AI's potential according to Goldfarb?

Goldfarb references Dario Amadei's essay, 'Machines of Loving Grace,' which envisions a future where AI could contribute to economic equality, healthcare improvement, and better governance. He believes that by involving real experts in AI development, we can start to see glimpses of this ideal future. As Forum AI tests and evaluates AI models with expert input, they are beginning to identify gaps and make improvements, paving the way for a more positive impact from AI technologies.