Data Protection in Traditional AI & Generative AI
07th July 2023
Hyderabad, India
Introduction

The nasscom Responsible AI Hub, in collaboration with Telangana AI Mission and Capgemini, hosted a multi-stakeholder roundtable on Data Protection in Traditional AI & Generative AI in Hyderabad on 7th July 2023.

The roundtable brought together leading technology, management, and regulatory experts to discuss and analyse emergent issues around data protection in AI development and use.

A list of roundtable participants is provided in the Annexure.

intorduction image
Key takeaways from the roundtable discussion
key takeaways
Participants discussed the need for all public and private data fiduciaries to uphold the imperatives of data privacy and data autonomy, regardless of the discriminative or generative nature of AI models.
Participants highlighted the privacy-preserving prospects of synthetic data generation and use for AI modelling.
Participants proposed optimising personal data collection and usage by narrowly defining the use case for a given AI model and the corresponding personal data requirements for building it (as opposed to the indiscriminate collection of personal data); participants highlighted that data privacy risks varied across industry sectors and use cases.
Participants emphasised the use of federated learning as an effective privacy-preserving technique for AI modelling.
Participants highlighted the need for global alignment on standards for data protection.
Participants discussed the possibility of data fiduciaries sharing benefits from the monetisation of personal data with data principals while stressing its importance to counter the concentration of profits in the hands of Big Tech.
Participants proposed companies to adopt a human-in-the-loop design as a measure to mitigate the risk of unauthorised data sharing and resulting privacy harm, while also fixing accountability for it.
Participants underlined the significance of free and open exchange of tools and best practices within developer communities to foster safe and responsible personal data management.
Participants highlighted the need to sensitise data principals on data privacy and related issues to ensure that the consent given for data sharing is informed and meaningful.
Participants proposed prompt filtering as a measure to pre-empt feeding of personally identifiable information to generative AI models.
ANNEXURE:
List of Participants
  • Anil Gurijala, Sr. Director of Engineering/General Manager, Fanatics
  • Ankit Bose, Head of AI, nasscom
  • Kaushik Rajan, Founder, Stoicus Legal
  • Kishore Seshagiri, Executive Director, Broadridge
  • Krishna Priya, Director Engineering, F5
  • Kothapalli Lakshman Kumar, Sr. Director, Infrastructure Architecture, S&P
  • Madhu Viswanathan, Associate Professor, Indian School of Business
  • Maunendra Desarkar, Associate Professor, Indian Institute of Technology Hyderabad
  • Rama Devi Lanka, Director-Emerging Technologies Wing, Govt of Telangana
  • Raviprasad Pisupati, Co-Founder & Partner, Tempus Law Associates
  • Ravi Srinivas, Senior Fellow & Consultant, NALSAR University of Law
  • Saurabh Saxena, Head of Engineering - Uber Operations Platform, Uber
  • Sibanjan Das, Senior Manager-Data Science, ServiceNow
annexure image
  • Srikanth Srinivasan, Vice President and Head, Membership and Outreach, nasscom
  • Srinivas Atreya, Chief Data Scientist, Cigniti
  • Sonu Kumar, AI Lead, Applied Innovation Exchange, Capgemini
  • Sudha Pelluri, Dean, Faculty of Informatics, Osmania University
  • Surya Prabha Vadlamani, VP-Enterprise AI & Cognitive Engineering, Centific
  • Venugopal Jarugumalli, DGM AI/ML, ZF