Until late last year, the term AI, for artificial intelligence, stirred little general interest. Online searches for the term had remained flat for years, and imagining, much less seriously assessing its potential impacts was mostly confined to futurist musings.
But then, at year’s end, came the launch of ChatGPT, described as “an artificial-intelligence chatbot” by OpenAI, the San Francisco company that created it. Once anyone could have a go at asking it questions or giving it tasks – to write an essay, create an image, or produce a piece of computer code – its possibilities and threats galvanised international discussion. Debate continues over whether such technology actually qualifies as artificial intelligence, when what it is doing is not a sign of sentience but ground-breakingly clever data processing and word associations. Still, terminology is a side issue to evaluating what this new, accessible and sophisticated technology means and the real world impacts it will have.
ChatGPT and similar AIs could transform work, by taking on a range of dreary basic content-production tasks, freeing workers to focus on more challenging, interesting and valued work. Or it could replace jobs, with devastating effect. And a decade of the roiling challenges and societal impacts of social media technologies have taught us to be wary of underestimating the knowns and the yet-to-be-imagined unknowns of transformative technologies, especially those released to general public use – and misuse.
The potential for harm was highlighted this week when Minister of State for cybersecurity Ossian Smyth tasked the National Cyber Security Centre with drawing up advice for citizens on the risks of this new generation of AIs being used for scams. International regulation for AIs is also pending. In September of last year, the European Commission issued a proposal for an AI directive. Regulation remains highly contested in an area so full of unknowns, and subject to vigorous industry lobbying. Yet the world must act, to help deliver AI’s promises while limiting its risks, especially its hidden but inevitable future threats.