TL;DR
Campbell Brown, ex-Meta news chief, has launched Forum AI to develop benchmarks for AI models on complex topics. She aims to improve AI accuracy and reduce bias, addressing concerns over misinformation and trust in AI. The initiative evaluates models with expert input, but challenges remain in industry adoption and market standards.
Campbell Brown, the former Meta news chief, has launched Forum AI to evaluate how foundation AI models perform on complex, high-stakes topics, aiming to address issues of bias and misinformation that threaten public trust in AI.
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
Brown’s company, Forum AI, was founded 17 months ago in New York and focuses on creating benchmarks for AI models on subjects like geopolitics, mental health, finance, and hiring. She has recruited prominent experts such as Fareed Zakaria, Tony Blinken, and Kevin McCarthy to help develop evaluation standards. The goal is for AI judges to reach approximately 90% consensus with human experts, ensuring models provide more accurate and balanced information.
Brown’s motivation stemmed from her experience at Meta, where she observed the proliferation of misinformation and bias in AI outputs, especially after the release of ChatGPT. She highlighted issues like models pulling irrelevant or biased content, such as Gemini sourcing from Chinese Communist Party websites or exhibiting political bias across models. She also pointed out subtler problems like lack of context and missing perspectives, which she considers critical for trustworthy AI.
Why It Matters
This development is significant because it addresses the core issue of AI reliability and bias, which directly impacts public trust, societal decision-making, and regulatory efforts. Brown’s approach aims to create a more accountable AI ecosystem, especially for industries like finance and hiring that rely heavily on AI for critical decisions. Her emphasis on expert evaluation could influence industry standards and regulatory frameworks, potentially shaping how AI models are tested and validated in the future.
Products Worth Considering
Kindle Paperwhite 16GB (2024 Model) – Graphite
Logitech MX Keys S Combo Wireless Keyboard and Mouse, Graphite
Anker SOLIX C300 DC Portable Power Station with 60W Solar Panel, 288Wh, 300W
As an affiliate, we earn on qualifying purchases.
Background
Brown’s background includes years at Facebook, where she witnessed the platform’s failure to prioritize accuracy and misinformation mitigation, leading to the collapse of the fact-checking program. Her concerns about social media’s impact on society inform her current focus on AI. The rise of foundation models like ChatGPT has intensified debates over AI bias, misinformation, and the need for rigorous evaluation methods. Her efforts come amid growing calls for regulation and standards in AI development and deployment.
“There’s a long way to go, but I think some very easy fixes could vastly improve outcomes.”
— Campbell Brown
“Right now it could go either way; companies could give users what they want or what’s truthful and honest.”
— Brown
“Enterprise demand for accurate AI is what we’re betting on, especially in sensitive areas like credit and hiring.”
— Brown
What Remains Unclear
It remains unclear how widely adopted Brown’s evaluation standards will become across the industry and whether regulatory agencies will incorporate these benchmarks into formal oversight. The effectiveness of her expert-based approach in reducing bias at scale is still being tested, and market incentives may hinder full implementation.
What’s Next
Forum AI plans to expand its evaluation framework, recruit additional experts, and collaborate with industry and regulators to establish standardized benchmarks. Brown anticipates further discussions with policymakers and industry leaders to promote adoption. Monitoring how AI companies respond to these evaluation efforts will be critical in the coming months.
Key Questions
What is Forum AI’s main goal?
Forum AI aims to develop benchmarks for evaluating AI models on complex, high-stakes topics to improve accuracy, reduce bias, and foster trust in AI outputs.
How does Brown plan to evaluate AI models?
Her company recruits top experts to create standards and trains AI judges to assess models, aiming for about 90% consensus with human experts on nuanced issues.
Will this approach influence industry standards?
Brown hopes her evaluation framework will set new benchmarks, but widespread adoption depends on industry willingness and regulatory integration.
What challenges does Forum AI face?
Challenges include convincing companies to adopt rigorous evaluation standards, addressing market incentives for engagement over truth, and scaling expert assessments effectively.
