Thrill-K: Rethinking knowledge layering and construction for higher machine cognition | Gadi Singer

The AI industry is now facing its next big challenge.

What are the necessary properties of representational structures that could allow vast amounts of data become meaningful in the human sense of the word?

How can knowledge architectures be constructed in a way that allows for both the efficiency and effectiveness of models they support?

In his Connected Data World 2021 keynote, Gadi Singer, VP & Director of Emergent AI at Intel Labs, discusses anthropomorphic conceptual structures and their benefits for enhancing Cognitive AI capabilities.

A visionary concept and a keynote which is even more timely today than it was then, foreseeing many current and, dare we say, future developments.

Singer introduces his model of the three levels of knowledge – Thrill-K – which can serve as a blueprint for building AI systems that are both efficient and scalable.

He begins with an Introduction to the Next Wave of AI. He addresses Language Models such as GPT-3, their shortcomings as Knowledge Models, and how they can be used in combination with Knowledge Graphs.

He then lists Five Essential Capabilities of Great Knowledge Models and describes the Thrill-K Architecture. Singer concludes by referring to The Future of AI, Cognitive AI and Deep Knowledge.

---

Connected Data London 2024 has been announced!.

December 11-13, etc Venues St. Paul’s, City of London

Check #CDL24 for more Presentations, Keynotes, Masterclasses, and Workshops on cutting-edge topics from industry leaders and innovators: https://connected-data.london