Aleksandr Tiulkanov on AI Policy, Laws, Regulation, and What We Really Need from Government – Voicebot Podcast 329
When you are scraping data from the internet, it is publicly available. Okay. But it was not necessarily initially published with the understanding it will be used in those particular scenarios to train AI systems. Maybe what you could be doing, as a responsible AI developer, you could be using those data but taking precautions like anonymizing data, pseudonymizing data, and taking other technical and organizational measures. If your purpose is to train a model, it doens’t matter whose data I use beause you are not gonig to target those precise people, but you still should make sure they are safe because you are combining a lot of data about people…If you started somewhere and people might be accessing it without permission, a lot of negative conseuqence could happen. But, if before you combine those data sets, you anonymize the data and ensure a particular person cannot be traced you are more on the safe side. There should be safeguards in place.
Google is not illegal. What they are doing is compliant in general with data protection regulations. I don’t see a big difference from what they are do crawling the internet to facilitate search from what people are doing with machine learning models.
Aleksandr Tuilkanov is a lawyer specializing in AI policy, regulation, and legal frameworks. Early work in GDPR and AI regulation before the generative AI frenzy led him to be hired by several governments to help draft AI policy. Few people have focused on AI policy exclusively for a significant period of time, and his insights are grounded in that experience. His work was influential at the Council of Europe, a key player in the latest EU AI regulatory framework.
We talk about imitation vs inappropriate use of data, appropriationism, whether chatbots are capable of defamation, copyright, IP protection, and what the real level of urgency is around regulating generative AI. We also talk about his recent colorfully-titled post, “Let’s not bomb the AI data centers just yet.”
Aleksandr Tiulkanov is an AI, data, and digital policy counsel. He earned his law degree at the University of Edinburgh and has specialized in AI regulation since 2015. He is a former senior manager at Deloitte, where he worked with corporations on AI policies. He later worked for several governments, including the Council of Europe, to design their AI policy and regulatory frameworks. He is currently a researcher at the Center for International Intellectual Property Studies and working independently with government agencies on AI policy.
Aleksandr Tiulkanov Interview – Show Notes
- Follow Aleksandr Tiulkanov on Twitter and LinkedIn: @shadbush Aleksandr Tiulkanov
- Webinar on AI policy by Aleksandr in June: Responsible Generative AI – Practical Guidance for Business
Follow @bretkinsella Follow @voicebotai
Adam Cheyer Co-founder of Siri and Viv Labs on Assistants, AI, and ChatGPT – Voicebot Podcast Ep 324
