Title: Language Models Solving Personal Data Interoperability
Speaker: Harri Ketamo, HeadAI
Data is everywhere but it is in silos, in different formats, standards, languages, contexts and so on. When consent to use the Text-based Personal Data is given, we are only in the beginning of the story. The quality of the results are dependent on all the steps during the data flow: quality of the data, quality of the methods in use and finally quality of the UX. Finally, the key is in context aware and conceptual interoperability, in which process ontologies are not enough, we must apply large language models in order to make data interoperable also as language.
Title: Personal Data Literacy in the Chat Era
Speaker: Johanna Walker, Elena Simperl
ChatGPT is a conversational agent built on top of OpenAI’s GPT3+ large language model. It recently became the fastest technology to reach 100m users, doing so within 2 months of launch. Given this reach, ‘chatting’ may become the new ‘googling’. Users are able to interact with ChatGPT through natural language conversations, and use this facility to compose a wide variety of text based works, including emails, essays and code. Using the right questions, or prompts, the same technology can also assist in data discovery.
The MyData declaration states, “We intend to help individuals have the tools, skills and assistance to transform their personal data into useful information, knowledge and autonomous decision-making”. We examine the implications of this in the ‘chat’ era, through asking the following questions:
Do conversational agents present an opportunity to support through acting as accessible data tools?
Do conversational agents present an opportunity to support this through acting as critical data literacy tools?
How can they be used to raise awareness of limitations in data and algorithms, and the use of personal data?
What do people need to know about how conversational agents might infer personal information about users in the future?
Title: Advancing AI Privacy: Enhancing Privacy Through MyData in Modern Generative AI Apps
Speaker: Robert Mao, ArcBlock
This talk dives into the rapidly evolving AI technology and discusses how emerging technologies such as Decentralized Identity (DID) and MyData are transforming the landscape of data privacy in generative AI applications. We will unravel the design of our decentralized computing model, illustrating how it underpins OpenAI and other AI models at the backend. By utilizing DID:Spaces, one of the latest MyData 2023 Operators, we demonstrate our proficiency in creating real-world applications that prioritize privacy.
Title: Large language models (LLMs) in Recruiting – A Powerplant or a Bomb?
Speaker: Ville Herva, CEO of Bolt.works
This session is hosted by Ulla Kruhse-Lehtonen.