CSET’s Helen Toner was cited by Time in an article about the European Union Artificial Intelligence Act, which, if passed, would make the EU the first major jurisdiction outside of China to pass targeted AI regulation.
In regards to the challenges with the EU AI Act and general purpose AI systems, Toner states, “The underlying problem here is that the whole way they structured the EU Act, years ago at this point, was by having risk categories for different uses of AI.” She further explains that large language models, such as GPT-4, do not have a specific use case and pose a significant shift in how AI works.
Toner emphasizes the difficulty in regulating such systems, explaining, “Once these models are trained, they’re not trained to do one specific thing. Even the people who create them don’t actually know what they can and can’t do. I expect that it’s going to be, probably, years before we really know all the things that GPT-4 can and can’t do. This presents a significant challenge for legislation that categorizes AI systems based on risk levels and their use case.”
Read the full article in Time.