Europe can make sure that AI is aligned with human values
Francesca Rossi is an IBM Fellow and IBM's AI Ethics Global Leader and one of the the keynote speakers at AI Day 2020.
Francesca Rossi has many positions in shaping AI. She co-chaires the IBM AI Ethics board and leads several of its work streams on AI ethics education, training, methodology, enforcing, operationalization, communication, and AI pipeline revision.
Rossi works with many institutions such as UN, WEF, IEEE, EC, to advance the governance and international collaboration on AI ethics. She has been the general chair of the AAAI 2020 conference and will be the AAAI President in 2022-2024. On top of all this, Rossi is a researcher.
What are you currently researching?
One of the projects I am leading is about getting inspiration from cognitive theories of human decision making, such as the “Thinking Fast and Slow” theory of Daniel Kahneman, to understand how to advance AI decision making capabilities.
Although very successful in many application domains, AI still lacks the ability to learn with few examples, to reason about causality, to rapidly adapt to new situations, and to generalize from raw data to high-level concepts that can be used for reasoning and communication to others. Humans have these capabilities. So the idea is to study what we know of human behavior and reasoning to understand the causes of our capabilities, in order to provide them in an AI scenario.This is a very multi-disciplinary project that involves IBM Research and several universities.
What has changed in AI research since you started your career?
Both AI capabilities and the AI research environments are fundamentally different. In the 80s and 90s, AI research was mostly done in academic environments. It also did not have access to huge amounts of data or large computing power, and was mostly symbolic and logic-based, rather than using data-driven approaches as in machine learning.
What and why ethical issues are important in the development of AI systems?
AI technology is very pervasive in our life, so it brings significant transformations. This can generate concerns and questions. The use of massive amounts of personal data, the opaqueness, the lack of explainability, the possible unfairness in the AI decisions, the impact on jobs and society, all these are AI features that require an ethics framework and governance to be handled in a way that prioritizes human well-being.
How does European approach to AI look like from the American view?
Europe is traditionally a place where technology is subject to more significant regulation compared to the US. However, these two regions of the world are aligned around a risk-based approach to AI applications, where only the high-risk applications require a mandatory regulatory approach.
How Europe should compete against large economies such as US and China?
Europe cannot and should not compete in terms of funding or other resources. However, it can be the place to show how an AI ethics approach to AI governance and regulation can support both innovation and competitiveness, and at the same time make sure that AI is aligned with human values.
What types of results you hope to witness in the next ten years in AI field?
I hope to see significant advances in AI learning and reasoning capabilities, which in turn will give us a deeper understanding of humans’ mind, intelligence, and values.
AI Day 2020 takes place on 26 November 2020. Read more at https://fcai.fi/ai-day-2020