A year of pondering on signals of change

Signal post, December 2020.

In April 2020, we launched this series of ETAIROS Signal posts to create space for pondering on signals of change – concrete events, shared experiences and current developments. These signals provide hints of what might be possible in the future and how we are shaping the future, especially in terms of AI technology, ethics and societal implications.

We started with AI ethics and handling of the unprecedented COVID-19 crisis. We asked what if it becomes acceptable to loosen or even lift AI ethics. This question actualized again in December 2020 with the news that the employment of Timnit Gebru with Google as technical co-lead of the Ethical Artificial Intelligence Team has been abruptly ended after she was allegedly forced to retract an academic paper, in which she highlighted risks from using large amounts of text data to train AI systems. This incident was followed by a lively discussion in social media, an open letter in support of Dr. Gebru, who is a Black woman, signed by 2,000 Google employees and 3,000 members of academia, industry, and civil society, as well as an apology for the controversy by Google CEO Sundar Pichai. Some voices have commented that this case raises troubling questions about corporate influence over AI research in general, and AI ethics in particular.

From the perspective of our research project, these events deserve attention because they indicate some tensions related to AI ethics that can trigger a societal response. What if in the future the ethical AI voices are seen acceptable by senior decision-makers only if they paint the company well? What factors affect dominant decision-makers and what is the role of various stakeholders in influencing these processes? What if critical voices become too loud and uncomfortable? What are our societal mechanisms for ensuring that these perspectives are not silenced? On the other hand, how can we know that voices promoting ethical AI are adequate and reasonable?

During this year, we tried to capture various signals related to recent developments and societal consequences of deploying AI technology in different contexts. Among the noise and buzz around AI development, we identified a clear signal pointing to the increasing emphasis on the need to understand the real consequences of AI. The 34thConference on Neural Information Processing Systems (NeurIPS) serves as a great example. The event was held virtually in December 2020 and it attracted more than 20,000 participants and featured interesting talks by leading experts in the AI field. In particular, we took notice of Georgia Tech professor Charles Isbell’s opening keynote, in which he suggested that it is time for the AI field to “grow up”, take responsibility for the real-world consequences of its work, and collaborate more actively with other stakeholders. Research is needed to study and support these activities, and further collaboration between various stakeholders. Foresight activities in ETAIROS project continue in 2021 with bringing different stakeholders to discuss the future of AI ethics and societal implications.

Hyvää Joulua ja Onnellista Uutta Vuotta! Merry Christmas and a Happy New Year 2021!


The writers: Santtu Lehtinen and Nadezhda Gotcheva (VTT)

Picture: Pixabay