Anthropic is stepping up its game in the AI world with a new initiative aimed at funding the development of comprehensive AI benchmarks.
Announced on Monday, this program will provide financial support to third-party organisations that can effectively measure advanced capabilities in AI models, including generative models like Anthropic’s own Claude.
The goal? To elevate the entire field of AI safety by creating high-quality, safety-relevant evaluations. Anthropic acknowledges the current benchmarking problem in AI, where existing benchmarks often fail to capture real-world usage and may not measure what they claim to, especially those created before the advent of modern generative AI.
Anthropic’s ambitious plan includes developing challenging benchmarks that focus on AI security and societal implications. These benchmarks will assess a model’s ability to perform tasks like cyberattacks, enhancing weapons of mass destruction, and manipulating people through deepfakes or misinformation. The company also aims to create an “early warning system” for identifying and assessing AI risks related to national security and defence.
Want to learn more about AI's impact on the world in general and property in particular? Join us on our next Webinar! Click here to register
In addition to security-focused benchmarks, Anthropic’s program will support research into AI’s potential for aiding scientific studies, conversing in multiple languages, and mitigating biases. They envision new platforms that allow experts to develop their own evaluations and conduct large-scale trials involving thousands of users.
Anthropic has hired a full-time coordinator for the program and is open to purchasing or expanding promising projects. The company offers a range of funding options tailored to each project’s needs, although specific details remain under wraps.
While Anthropic’s effort is commendable, there’s some scepticism about its commercial ambitions and the influence it may exert on defining “safe” or “risky” AI. Critics argue that focusing on catastrophic AI risks, like nuclear weapons, may divert attention from more immediate regulatory issues, such as AI’s tendency to produce hallucinations.
Anthropic hopes this program will set a new industry standard for comprehensive AI evaluation. Whether the broader AI community will embrace this initiative, given Anthropic’s commercial interests, remains to be seen.
Stay tuned for more updates on this groundbreaking program and its impact on the AI landscape. Keep an eye on our blog for the latest in AI advancements and industry insights!
Want to learn more about AI's impact on the world in general and property in particular? Join us on our next Webinar! Click here to register
Made with TRUST_AI - see the Charter: https://www.modelprop.co.uk/trust-ai
Kommentit