AI Vulnerability Database

? AVID: 2023 in review As AI became a household topic in 2023, the need for rigorous approaches to managing the risks of this emergent technology became obvious to the general public, companies, and governments around the world. AVID provided an outlet to this groundswell of interest. During the rest of this year, we launched a slew of public events to channel this interest to productive outlets, partnered with like-minded organizations in community efforts, and had a number of releases seeding technical resources that would be indispensable to AI risk practitioners for decades to come. As a purely grassroots organization, we scaled unfathomable peaks during 2023. We've partnered with AI Village to organize the biggest ever AI red teaming event, organized a number of community events, won grants, seeded developer tooling, and published research on AI safety and risk management. We couldn't have done without our community and network of collaborators spanning the globe. To all our friends, thank you for your support! Read more here: https://lnkd.in/grasRD8C We wish everyone Happy Holidays. See you all in 2024!

Check out this new podcast, from our very own Jekaterina Novikova, PhD, to learn more about women making waves in AI research. Knowing her, it proves to be both deeply informative and challenging.

We're excited for the outputs from this collaboration between Data & Society Research Institute and ARVA taking shape. We hope you can join this event to hear from the team, including our board member Borhane Blili-Hamelin, PhD!!

Our Science Lead Jekaterina Novikova, PhD will part of this panel discussing efficient LLMs tomorrow. Be there!

Join us tomorrow!

I'm looking forward to participating in this event tomorrow, where we'll talk about end-user audits of AI systems. Join us! #aiethics #responsibleai

The power to uncover harmful flaws in AI shouldn't rest solely with experts. But how can everyday users contribute to responsible AI development? Join Michelle Lam, Christina Pan, Carol Anderson, Nathan Butters, and Borhane Blili-Hamelin, PhD to explore IndieLabel, an innovative tool that puts algorithmic audits in the hands of everyday users. Come discuss the power of public participation in shaping the future of AI. Speakers Michelle S. Lam, Christina A. Pan, Carol Anderson, Nathan Butters, and Borhane Blili-Hamelin Date & Time Wednesday July 24, noon to 1pm ET.

Interested in the problem of expanding participation in AI evaluation? What other innovative approaches can empower the public to hold AI accountable? How can everyday users contribute to responsible AI development? Join us for this conversation!

Join Jekaterina Novikova, PhD Novikova, Science Lead at AI Vulnerability Database! Explore the?"Dual Nature of Consistency in Foundation Models: Challenges and Opportunities"?in her upcoming talk. ???? What You'll Learn: - Understanding Consistency:?Gain insights into why consistency is crucial for trustworthy LLMs and foundation models. - Measuring Consistency:?Learn how to evaluate consistency in these models effectively. - Mitigation Practices:?Discover strategies to address the negative consequences of inconsistencies. - Leveraging Inconsistencies:?Find out how observed inconsistencies can be turned to your advantage with practical examples. Why Attend? - In-Depth Knowledge:?Benefit from Jekaterina's expertise on the complex issue of consistency in AI models. - Practical Strategies:?Learn actionable mitigation practices to enhance model reliability. - Innovative Examples:?Explore innovative ways to leverage inconsistencies for positive outcomes. About the Speaker: Jekaterina Novikova?is the Science Lead at the AI Risk and Vulnerability Alliance. She specializes in understanding and addressing the risks and vulnerabilities associated with AI models, particularly focusing on consistency in foundation models. ???Don’t miss this opportunity!?Secure your spot now: https://lnkd.in/g6EEsEc Join Jekaterina Novikova for a comprehensive exploration of consistency in foundation models!

?? Thrilled to announce that our paper, "Closing the Loop: Embedding Observability in the GenAI Product Lifecycle for Systematic Bias Mitigation," has been accepted under the "Harms: Privacy" theme at Generative AI and HCI (GenAICHI) at ACM #CHI2024 in Honolulu, Hawaii! Under Nimmi Rangaswamy's guidance, I explored the transformative potential of the Observability and Governance Layer within AI systems. This layer acts like a high-powered microscope, zooming in on AI operations, ensuring adherence to strict security protocols, and providing a feedback loop that sends critical insights back to the core of AI development. This continuous cycle of feedback and improvement enhances the overall integrity and fairness of AI technologies. ?? Paper: https://lnkd.in/dRkw6mJH A cornerstone of our approach is the BiasAware initiative, inspired by my collaboration with the AI Vulnerability Database last year. This highlights the strength of open and trusted platforms in enhancing our methodologies and broadening the impact of responsible AI practices. It's important to recognize that AI is an evolving field. As we continue to refine our understanding and implementation of observability, our work with the GenAI product lifecycle aims to lead the way in creating AI systems that are not only intelligent but also responsible and equitable. If you’re at CHI, come join me in Room 323A at GenAICHI to dive deeper into these discussions! #GenAICHI #CHI2024 #EthicalAI #Observability #AIResearch