TL;DR
Campbell Brown, ex-Meta news chief, launched Forum AI to develop benchmarks for evaluating AI on high-stakes topics. She highlights concerns about current AI biases and the need for better accuracy. The initiative aims to influence how AI delivers truthful information.
Campbell Brown, a former Meta news chief, revealed her new company, Forum AI, which aims to develop benchmarks for evaluating AI models on high-stakes, nuanced topics such as geopolitics and mental health, to improve accuracy and reduce bias.
Brown founded Forum AI 17 months ago in New York after witnessing the rise of ChatGPT and realizing the potential risks of misinformation. Her company evaluates foundation models on complex issues by collaborating with top experts like Fareed Zakaria, Tony Blinken, and others to create benchmarks and train AI judges to reach about 90% consensus with human experts. Brown noted that current models, including Gemini, show biases, such as pulling from Chinese Communist Party sources or leaning politically, and often miss important context or perspectives.
Brown emphasized that her motivation stems from her experience at Meta, where she observed how platform optimization for engagement led to societal misinformation and polarization. She advocates for AI to prioritize truth, especially in enterprise applications like credit and hiring, where liability demands accuracy. However, she criticizes current AI compliance measures, citing examples like New York City’s biased hiring audits that often miss violations, underscoring the need for domain expertise and thorough evaluation.
Why It Matters
This matters because it addresses the core challenge of ensuring AI delivers truthful, unbiased information, which impacts public trust, societal discourse, and enterprise accountability. Brown’s efforts could influence how AI models are tested and regulated, potentially leading to more responsible AI deployment and reducing misinformation.
Jeimier 5 Sizes Bias Tape Makers, Upgraded Bias Binding Tape Making Tool for Fabric Quilting Sewing, Quickly Customize, Solidly Bias Quilting Tool, 1/4IN 3/8IN 1/2IN 3/4IN 1IN
QUICKLY MAKE BIAS BINDING: The Jeimier 5 sizes professional Bias Tape Makers out of any fabric to match…
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Background
Brown’s background includes years at Facebook and Meta, where she observed the pitfalls of optimizing social platforms for engagement. Her departure and subsequent founding of Forum AI reflect a growing industry concern over AI’s role in misinformation, bias, and the need for rigorous evaluation standards. The industry has faced criticism for inadequate compliance and transparency, especially in high-stakes applications like hiring and finance.
“There’s a long way to go, but I also think that there are some very easy fixes that would vastly improve the outcomes.”
— Campbell Brown
“Right now it could go either way; companies could give users what they want, or they could give people what’s real and what’s honest and what’s truthful.”
— Brown
“The compliance landscape is a joke. Real evaluation requires domain expertise to work through known scenarios and edge cases.”
— Brown
AI accuracy benchmarking software
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
What Remains Unclear
It is still unclear how widely Forum AI’s benchmarks will be adopted by AI developers or regulators, and whether industry giants will prioritize transparency and accuracy over commercial interests. The effectiveness of AI judges in reaching consensus on complex topics remains to be validated at scale.
Agentic Artificial Intelligence: Harnessing AI Agents to Reinvent Business, Work and Life
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
What’s Next
Forum AI plans to expand its benchmarking efforts, collaborate with more industry partners, and potentially influence regulatory standards. The next milestones include publishing detailed evaluation results and advocating for broader adoption of rigorous testing protocols.
high-stakes AI testing platforms
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
What is Forum AI’s main goal?
Forum AI aims to develop benchmarks and evaluation methods to improve AI accuracy and reduce bias on complex, high-stakes topics.
How does Brown propose to improve AI trustworthiness?
By collaborating with experts to create standards and training AI judges to evaluate models, aiming for high consensus with human experts on nuanced issues.
Will industry giants adopt these benchmarks?
It remains uncertain. Adoption depends on regulatory pressures, industry willingness, and demonstrated effectiveness of the evaluation methods.
Why is this effort important now?
As AI becomes central to information dissemination and decision-making, ensuring its accuracy and fairness is crucial to prevent misinformation and bias.
