The Dawn Project, a safety advocacy group, is intensifying its efforts to highlight the shortcomings of artificial intelligence (AI) systems and the risks associated with deploying them in safety critical scenarios.
The group is aiming to raise awareness about the potential dangers of catastrophic cyber attacks resulting from the use of easily hackable, flawed commercial-grade software in critical applications like water systems and the power grid. Despite clear flaws in AI systems, such as chatbots failing to answer basic questions or inaccurately rewriting information, The Dawn Project argues that AI developers are exaggerating the progress made by these systems and pushing for their deployment in crucial infrastructure worldwide.
As part of their campaign, which includes advertising in major publications like The Wall Street Journal and The New York Times, Dan O’Dowd, founder of The Dawn Project, criticized Tesla’s AI-powered Full Self-Driving software. He cited incidents where the technology illegally passed a stopped school bus and endangered children crossing the road.
The group also drew a comparison between AI proponents and deceptive salespeople, warning against blindly trusting these technologies for critical functions that could lead to catastrophic consequences. They emphasized the need for consistency and reliability in safety critical systems, advocating for stringent software security standards to be applied to infrastructure vital to millions of lives.
The Dawn Project firmly stated that commercial-grade software is not suitable for applications where human harm is a possibility, urging caution against relying on AI systems that have already shown their limitations. They questioned the readiness of these technologies to be used in potentially life-threatening scenarios, stressing the importance of prioritizing safety and reliability in critical infrastructure solutions.