Cornell University researchers received a $250,000 seed grant and a chance at a $10 million award from the Rhode Institute for support 5-year “Moonshot” project Using artificial intelligence, we aim to establish a new foundation for trusted AI-mediated communication across online platforms.
As AI agents increasingly influence our online conversations and shape the way information is presented, interpreted, and discussed, the cross-university team strives to ensure that this influence is transparent, independently verifiable, and consistent with the public interest. The team will develop tools and protocols that allow platforms and users to see and validate how AI systems shape conversations. The long-term vision is a new communication ecosystem where AI can support and mediate discussions at scale.
“The online experience is being reshaped in almost every dimension with the advent of AI-powered conversational agents,” said the lead researcher. John Kleinberg 1993, Professor of Computer Science and Information Science, Cornell Ann S. Bowers College of Computing and Information Science, Tisch University. Already, someone can interact with a person online using an AI writing assistant, bot, or AI agent acting on their behalf, with the interaction managed behind the scenes by an AI moderator, he said.
“Our shared public realm is currently undergoing a crisis of trust,” said the co-investigators. Cristian Danescu-Niculescu-MigillCornell Bowers Associate Professor of Information Science. “People no longer believe what they read, they don’t know who they’re talking to, and perhaps more importantly, they don’t trust the AI systems that mediate their interactions.”
By combining cryptography-inspired verification techniques with insights from communication and social science theory, the team develops the technical and social infrastructure for a trusted civil discourse environment. The goal is not to have everyone agree or trust each other, but to ensure that the environment itself supports meaningful and trustworthy interactions, even in the presence of AI.
Researchers liken this vision to a ride-sharing platform. People routinely ride in cars driven by strangers. This is because the system provides mechanisms to make interactions verifiable and reliable.
In addition to Mr. Kleinberg and Mr. Danescu-Niculescu-Migill, the following co-researchers are: david rand 2004, Cornell Bowers, Professor, Cornell SC Johnson College of Business, College of Arts and Sciences. Natalie BazarovaMA’05, Ph.D. In 2009, he became Associate Professor of Research and Professor of Communication in the College of Agriculture and Life Sciences. Robert D. KleinbergCornell Bowers Professor of Computer Science. and mor naamanthe Don Follett and Mibbs Follett Professors of Information Science at Cornell Tech, Jacobs Technion Cornell Research Institute, and Cornell Bowers.
In addition to AI assistants that support individual participants and AI mediators that guide group discussions, this project introduces entirely new civic building blocks. The AI audit layer generates a transparent record of each agent’s intervention, allowing independent verification of whether the agent’s actions are in line with its declared role and standards. These components work together to establish a new foundation for online communication. With this foundation, AI can actively participate in discussions while remaining transparent and independently verified.
“We have shown that AI models can be highly persuasive and useful, such as debunking conspiracies, and can be manipulative, such as when swaying political opinion,” Rand said. “To protect people without sacrificing the benefits of AI, we are building an auditing infrastructure that reveals the goals and intent of models and allows users to distinguish between useful and manipulative models.”
This seed grant is one of eight awarded by the Lord Institute as part of the Moonshot Research Competition, which invites researchers at the forefront of AI to tackle some of humanity’s toughest problems. Over the next six months, eight teams will compete to develop initial products and a full range of proposals for the $10 million Moonshot Lab, with winners selected later this year.
Moonshots is a flagship initiative of the Laude Institute, founded by Andy Konwinski, a computer scientist who co-founded Databricks, a cloud-based platform for data analytics, and Perplexity, an AI search engine, to accelerate the transition of cutting-edge computer science research to real-world impact. The Moonshots program is chaired by Dave Patterson, professor emeritus of computer science at the University of California, Berkeley, and a leading expert in the field.
“Moonshot was built on a simple premise: The world’s most important AI researchers should be the ones deciding how AI is used, and they deserve the resources to think at the largest possible scale,” Patterson said. “The results from 600 researchers from 47 institutions exceeded everything we expected. This is what open academic research looks like, where we can be very ambitious.”
“The Lord Institute’s innovative funding model is exactly the forward-thinking approach needed to harness the potential of AI for the public good,” said Chancellor Kavita Bala, professor of computer science at Cornell Bowers College. “Their support of this project, which embodies Cornell’s commitment to addressing society’s most pressing challenges, will transform the team’s visionary ideas into AI that makes us more human.”
During the first six months, the team will develop a conversational arena and framework to evaluate and compare AI assistants and mediators. These were developed by both internal and external researchers and partners. The evaluation progresses in stages from simulated interactions to real human conversations.
In addition, a dedicated “adversary track” has been added to stress-test the audit layer, inviting AI agents designed to simulate covert attempts to manipulate discussions while only appearing to comply with their publicly declared roles.
“If successful, this will fundamentally change the way we humans communicate in the digital realm, enabling more meaningful, constructive and trusting dialogue even when we have deep disagreements,” said Danescu-Niculescu-Migill. “The six-month turnaround time is very short, but that’s what makes it so appealing. It allows us to bring together teams with complementary expertise who might not otherwise have had the opportunity to work together, and to jointly focus our efforts and attention on one ambitious project.”
Rene KizilczekThe Cornell Bowers associate professor of information science leads a team aimed at helping workers transition into new roles in the AI economy by using AI to transform job roles into practice activities for job seekers ahead of interviews. Additional team members include: Michelle BelloFrances Perkins Professor of Industrial Labor Relations and Professor of Economics in the ILR School and College of Arts and Sciences. Torsten JoachimJacob Gould Schulman Professor in the Cornell Bowers Department of Computer Science and Information Science. JR KellerAssociate Professor of Human Resources at ILR school; philip kircherthe Irving M. Ives Professor of Industrial Labor Relations in the ILR School. Rachel Slama, Associate Director Future of Learning Lab Located at Cornell Bowers.
Rachi SinghThe Cornell Bowers assistant professor of computer science will lead a second group proposing the development of an AI agent that can train surgeons and assist them during remote telesurgical procedures. Her team includes: Emad Manzoorassistant professor of marketing at SC Johnson University. Misha Doerer of Ericsson. Dr. Ryan Madder, cardiologist at Corewell Health;
Patricia Waldron is a writer in the Cornell Ann S. Bowers College of Computing and Information Sciences.
