Google's DeepMind Forms New Research Group For AI Morality

Advertisement
Advertisement

AlphaGo creator and Google subsidiary DeepMind has announced the formation of the DeepMind Ethics & Society (DMES) group designated to researching and better understanding the impact of AI on the human society. The research group will be focused primarily on exploring the key ethical challenges facing the field of AI, such as AI morality and values, the impact automation might have on the economy, and AI management risks.

According to DeepMind, a better understanding of the social and ethical implications of AI requires DMES to be governed by five principles that should guarantee the transparency and social accountability of its work. These five core principles will be based on the social benefits of AI technologies, rigorous and evidence-based research conformed to the highest academic standards, as well as transparency and openness in regards to the projects the company will fund. The group is also committed to supporting public and academic dialogues revolving around the concept of AI, and to involve a broad range of viewpoints into its work as questions raised by artificial intelligence extend well beyond the technical field.

DMES presently entails eight full-time staff members while the company is planning to increase that figure to around 25 within the next year. The group also has six independent advisors called "Fellows" who will help the company in guiding its research strategy and work program. The list includes Oxford University Professor and AI expert Nick Bostrom, University of Manchester Professor Diane Coyle, and Edward W. Felten of the Princeton University. DMES researchers won't have a direct saying in the creation of DeepMind products but will instead work in parallel with the company's development teams to ensure that its work will be focused on the most urgent, difficult, and sensitive ethical challenges posed by AI. DeepMind and DMES will start publishing research on those AI-centered topics in early 2018, Google's subsidiary confirmed, without providing a more specific time frame for its upcoming efforts. The wording of the firm's latest announcement suggests that DMES is meant to be a long-term initiative that will accompany its research and development endeavors as the AI technology continues maturing going forward.

Advertisement