ability to evaluate the performance and the impact that artificial intelligence has and use the data received.
Antrophic is an AI research company whose prime purpose is to find a way in which artificial intelligence can function safely. The company is based in San Francisco and is formed out of teams with experience across ML, physics, products, and more. The purpose is to discover the right techniques and practices to keep using AI safely.
In order to do so, the AI safety company Anthropic will launch a program that has meaning to fund the development of those new benchmarks. Even more so, the program will be able to allocate payments to third-party organizations that can help sustain their cause.
“A robust, third-party evaluation ecosystem is essential for assessing AI capabilities and risks, but the current evaluation landscape is limited. Developing high-quality, safety-relevant evaluations remains challenging, and the demand is outpacing the supply,” reported Anthropic in a blog post on their website.
Even more so, the company has a benchmarking shortcoming – the manner in which AI is actually used by daily users. How do they interact with the software and how can we make sure that they are tested appropriately.
Anthropic is proposing a solution that will work toward the highest priority areas such as the AI safety level assessments, for advanced capability and safety metrics, and even for “Infrastructure, tools, and methods for developing evaluations”. Those tests and the ability to accomplish a higher level of protection for all users and minimize the risks for everyone.
“We offer a range of funding options tailored to the needs and stage of each project,” writes Anthropic in a blog post from their blog, further stating that “Teams will have the opportunity to interact directly with Anthropic’s domain experts from the frontier red team, fine-tuning, trust and safety, and other relevant teams.”.
In their blog post, Anthropic writes that they hope that they will be able to be “a catalyst for progress towards a future where comprehensive AI evaluation is an industry standard.”, hope for a better future where AI can not cause harm or any problems for our society.