Letter: Consumers have a right to know when they’re talking to a chatbot

The concern around Google’s chatbot, LaMDA – a Language Model for Dialogue Applications – and the perception that it might have become sentient reflects society’s deep obsession with powerful artificial intelligence (“Google puts on leave engineer who claimed bot is ‘sentient'”, Report, June 14).

It probably speaks more to human anthropomorphism than the inherent capabilities of the technology. Artificial intelligence and machine learning are immensely capable technologies but data scientists and ethicists have agreed (with notable exceptions in the current instance of course!) That the holy grail of AGI or “artificial general intelligence” has not yet been reached. That LaMDA can contextualize its statements so marvelously is a testament to both the depth and reach of the data sets it has reviewed and the deployment of incredibly complex mathematics – but let us be clear, LaMDA cannot think.

What this lesson teaches us is that, in this environment of convincing “fake” machine conversationalists, it is vital that AI is regulated to ensure transparency. An individual should have a right to know when they are interacting with an AI machine, not a human.

The EU is well on its way to a universal obligation to inform customers that they are dealing with an AI system. The draft Artificial Intelligence Act (or AIA) specifically deals with this problem in its transparency provisions. While the AIA is not perfect, it at least presents a solution in this regard.

In the UK we are still waiting for a delayed white paper and policy remains in development. There is a risk of a consumer backlash, or worse, if we do not declare to users at the point of use that they are dealing with machines. Such an approach will help to foster trust in the use and deployment of AI technology.

The launch of the UK Digital Strategy last week spoke of light-touch AI regulation, but was entirely missing the sense of urgency that this field increasingly needs.

The foundations of the UK’s world-class AI industry will ultimately be undermined if we fail to address this problem directly.

John C Buyers
Partner, Head of AI
Osborne Clarke LLP, London EC2, UK

Leave a Comment

Your email address will not be published.