The availability of large troves of personal data has seen the re-emergence of AI. With unfettered access to enormous amount of data, intelligent-systems are getting smarter and more accurate by the hour. With such rapid success, we ought to question the control user possess in terms of their expectations of privacy and accessibility to their own data.
In this session, we discuss the emerging technologies which advocate for alternative user-empowering paths to processing and accessing personal data, in contrast to the current standards of siloed data-accumulation and centralized data-processing. Furthermore, we touch upon the human intuition for privacy, largely seen in spoken communication, and why we should build smart machines modeled with similar intuitions in order to address user-centric privacy expectations.
Presentations in this session:
- Pluralism, Computational Cognitive Modeling and Human-empowerment in the Digital World by Soheil Human
This talk proposes that legal acts, like the GDPR, should be accompanied by the development of complementary technical solutions such as Cognitive Personal Assistant Systems to support people to effectively manage their personal data processing on the Internet. Considering the importance and sensitivity of personal data processing, such assistant systems should not only consider their owner’s needs and values, but also be transparent, accountable and controllable.
Pluralist approaches in computational cognitive modeling of human needs and values which are not bound to traditional paradigmatic borders such as cognitivism, connectionism, enactivism, etc., I argue, can create a balance between practicality and usefulness, on the one hand, and transparency, accountability, and control, on the other, while supporting and empowering humans in the digital world. Considering the threat to digital privacy as significant to contemporary democracies, the future implementation of such pluralist models could contribute to power-balance, fairness and inclusion in our societies.
- Hottest Trends in Privacy-Preserving AI : Federated Learning, AI on Encrypted Data and Membership Inference Attacks by Oguzhan Gencoglu
Current implementations of machine learning and AI algorithms require access to data, often centralized, which raises potential security and privacy concerns. Recent advancements in AI research introduces promising opportunities in privacy-preserving machine learning algorithms including training of AI systems without a centralized database (federated learning), AI that can run on encrypted data and using AI to predict if ones data has been used for training a black-box AI API (membership inference). This talk will provide a non-technical dive into the implications of such advancements by providing several real-life examples. The main objective of this talk is to raise awareness on such cutting-edge technologies and consequently, to spark up fruitful discussions around the future of privacy-aware AI and MyData paradigm which will have a substantial impact on society as well as business, possibly sooner than later.
- With whom are you talking? Privacy in Speech Interfaces by Tom Bäckström
Speech is about interaction. It is more than just passing messages – the listener nods and finishes the sentence for you. Interaction is so essentially a part of normal speech, that non-interactive speech has its own name: monologue. Normal speech is about interaction. From the myData-perspective, this means that all (speech) data is about interactions, between two or more parties. The interactive component makes speech more interesting and this applies to other data as well. For example, a photo of a beach becomes more interesting when you know that I went there with Sophie. Ownership of such data is then also shared among the participating parties. There is no singular owner of data, but access and management of data must always happen in mutual agreement.In our view, data becomes more interesting when it is about an interaction. To include the most significant data, we should therefore turn our attention from myData to focus on ourData. Here, the importance of data is then dependent on, and even defined by, with whom are you talking?