Should we need a licence to use AI?
In May 2023, the Center for AI Safety (CAIS), a non-profit organisation whose mission is to reduce societal-scale risks from artificial intelligence, published the following Global Statement on AI Risk signed by 600 leading AI researchers and public figures:
Mitigating the risk of extinction from AI should be a global priority alongside other societal-scale risks such as pandemics and nuclear war
Two months earlier, many of the same individuals signed an Open Letter published by the Future of Life Institute calling on all AI labs to immediately pause for at least 6 months the training of AI systems more powerful than GPT-4. It didn’t happen.
The response to these pleas has been for some governmental organisations to produce, but not yet implement, legislation to exercise some degree of control over the development of AI products.
Are legislators going far enough in controlling AI product development? Should they also be placing controls on users and usage of these products? While this may seem a step too far, experience from the regulation of the use of automobiles could provide some useful pointers.
In most countries, to drive an automobile you need to pass a test to obtain a licence which may be suspended if you fail to observe several regulations, such as speed limits, traffic lights, drink driving laws and the use of seat belts. Could the same principles be applied to using AI products? In this scenario, users of certain AI products, for example generative AI products such as ChatGPT, would require a licence which could be withdrawn if they fail to follow an established set of regulations.
To understand the potential opportunities of such an approach, it's informative to understand how long and difficult it was to introduce regulations and controls applied to driving automobiles. Vested interests played a significant role in resisting a number of the developments on the grounds that they impacted personal freedom but they are now considered to be sensible and acceptable.
The table below provides a timeline of some of the key developments.
The history of speed limits is a little more complex.
For example, in the United Kingdom, at the time of the first long-distance car journey in 1895, there was a speed limit on UK roads of 2 mph in a populated area and 4 mph elsewhere. In 1896, following lobbying from car manufacturers and motoring enthusiasts the maximum speed limit was increased to 14 mph. This development is still celebrated with the annual London to Brighton veteran car run. The speed limit was increased in 1903 to 20 mph but was widely ignored and was abolished in 1930. A 30-mph limit in "a built-up area" was introduced in 1936 and a 70-mph limit on motorways in 1965.
The experience of controlling the use of automobiles highlights the challenges in implementing similar controls but also demonstrates that applying similar controls to other technologies may be both possible and acceptable.
Jim Foster, Director, CMCE
