Ensuring artificial intelligence benefits humanity

22 January 2015

I Stock_000004797569Medium
© iStock

The past year has seen significant commercial investment in artificial intelligence research, including Google’s £400 million acquisition of London-based start-up DeepMind. But how can we ensure that developments in artificial intelligence will benefit, rather than harm, societies?

Daniel Dewey, Alexander Tamas Research Fellow at the Oxford Martin Programme on the Impacts of Future Technology, is helping to shape global AI research priorities as one of the principal authors of a new set of guidelines, published following a gathering of the world’s foremost AI researchers at the start of 2015.

The document, ‘Research priorities for robust and beneficial artificial intelligence’, was drafted with Stuart Russell (University of California, Berkeley) and Max Tegmark (Massachusetts Institute of Technology), and released by the Boston-based Future of Life Institute (FLI), which convened the conference. The FLI at the same time published an open letter outlining the need for AI to be beneficial to humanity, the signatories of which included Stephen Hawking, Lord Rees, the Oxford Martin School’s Professor Nick Bostrom. Elon Musk, founder of SpaceX, Tesla Motors and Paypal, also put his name to the letter, and announced he would fund a $10 million global grant programme for research on the future impacts and safety of artificial intelligence.

Daniel, who is based at Oxford’s Future of Humanity Institute, established in 2005 with funding from the Oxford Martin School, said he was pleased to see long-term consideration of AI’s impacts being brought to the fore.

“It is very exciting to see research of the type that the Future of Humanity Institute and its funders and supporters have been pushing for so hard - through Nick Bostrom's book Superintelligence, the Oxford Martin School Programme on the Impacts of Future Technology, the Alexander Tamas Fellowship, and FHI's core research programme - brought to greater attention and acceptance in the AI community”, he said. “We hope that this will mark a noticeable step toward a better long-term future, and are grateful to have been able to contribute to it.”

The conference also saw Professor Bostrom give a talk entitled ‘The Road Ahead’, and a presentation by Dr Michael Osborne on the impact of automation on employment, which forms a major part of the work of the new Oxford Martin Programme on Economic & Technological Change.