The UK has debuted what it calls a groundbreaking toolset for safety testing of artificial intelligence (AI).
The new proposal is announced Friday (May 10th) AI Safety Research Institutea software library called “Inspect” that can be tested by testers such as startups and academics. and International government AI developers evaluate specific features of individual AI models and generate scores based on their results.
Inspect is the first AI safety testing platform overseen by a state-sponsored agency and released for broader use, according to a news release from the institute.
“As part of the constant drumbeat of UK leadership on AI safety, I have cleared the AI Safety Institute’s inspection platform (called Inspect). open source,” Said Michelle DonnellanUK Secretary of State for Science and Innovation and technology.
“This will put British ingenuity at the heart of the global effort to make AI safe and solidify our position as world leaders in this field.”
The announcement comes a little more than a month after the UK and US governments pledged to work together on secure AI development. agree to cooperate About testing cutting-edge AI models.
“AI continues to develop rapidly, and both governments recognize the need to act now to ensure a common approach to AI safety that addresses the technology's emerging risks.” . US Department of Commerce he said at the time.
The two countries have also agreed to partner with other countries to promote AI safety around the world, holding at least one joint test in a publicly accessible model and establishing a The plan is to leverage collective expertise by exploring exchanges. .
This partnership is AI Safety Summit At the meeting, held last November, world leaders considered the need for global cooperation to combat potential risks associated with AI technologies.
“This new partnership means more responsibility will be placed on companies to ensure their products are safe, reliable and ethical,” AI Ethics Evangelist andrew perry A global intelligent automation company abby spoke to PYMNTS shortly after the collaboration was announced.
“Disruptive technology innovators tend to release products with a ‘ship first, fix later’ mentality to gain first-mover advantage. For example, OpenAI has provided some transparency about the potential risks of ChatGPT. I have, but Released for widespread commercial use Despite the harmful effects. ”
For all of our coverage of PYMNTS AI, subscribe to our daily subscription AI Newsletter.