Andrew Bailey says regulating AI is not the right place to start
Andrew Bailey writes on Substack that AI risks are real but development must not be halted, arguing regulation should not be the first step.
Stock photo for illustration only, not from the actual event
- Andrew Bailey states AI risks are real but opposes halting development.
- Argues that immediate heavy regulation is not the right starting point.
- Highlights UK testing efforts through the AI Security Institute.
- Emphasizes that unexpected model behavior proves testing is necessary.
Writing his very first article on Substack, Andrew Bailey stated that the risks surrounding artificial intelligence are real and increasingly significant. He emphasized that the development of AI should not be halted or prohibited because the benefits are immense, though there must be a system for intervention and clear operating boundaries.
In recent weeks, debates regarding AI risks and human implications have intensified. Bosses of leading AI firms such as Anthropic and OpenAI previously called for slower development and a coordinated international approach. OpenAI recently announced it would withhold its newest model due to safety concerns.
Stock photo for illustration only, not from the actual event
However, the idea of slowing down has faced rejection from figures like President Donald Trump, who stated that the US leads China in the AI race and added that whoever wins AI wins it all.
"That is not evidence that testing has failed, rather it is evidence of why testing is necessary."
Andrew Bailey
In the UK, Bailey noted that important testing work is underway via the AI Security Institute to help form standards. He acknowledged that testing will involve failures and models behaving unexpectedly, but argued this demonstrates the absolute necessity of rigorous testing.
Bailey's perspective highlights a pragmatic approach to balancing technological innovation with risk management. By opposing a blanket ban or premature heavy regulation, he advocates for a framework built on empirical testing and gradual standard-setting rather than stifling technological advancement from the outset.
Bailey concluded that testing should not be viewed as an alternative to future regulation, as a formal regulatory framework may emerge over time. Nevertheless, in his view, strict regulation is simply not the right place to start.
Source: BBC Business
Found something wrong in this article? Report an issue with this article
Comments
Leave a Comment