Magnified with Matt Cooper AI Magnified EP1 - Risks And Opportunities w/Patricia Scanlon and John Clancy
Oct 16, 2024
Patricia Scanlon, Ireland's AI Ambassador and founder of Soapbox Labs, teams up with John Clancy, founder of Galvia, to explore the evolving role of AI. They dive into the distinction between narrow and general AI, and how it influences our daily services like Netflix and Spotify. Discussion turns to generative AI, addressing concerns about data accuracy and bias. They also highlight the innovative use of AI in education with tools like the student assistant Cara, and the regulatory landscape’s impact on startups. A fascinating look at AI’s future in Ireland!
AI Snips
Chapters
Transcript
Episode notes
Narrow Versus General Intelligence
- AI excels in narrow, data-rich tasks but is not broadly intelligent like humans.
- Large models outperform humans in specific domains by ingesting far more data than any person can.
What Makes LLMs Powerful (And Costly)
- Large language models (LLMs) are foundation models trained on massive internet corpora and can be fine-tuned to company data.
- Their size and inference costs make them powerful but expensive to run at scale.
AI's Hidden Energy And Location Costs
- AI inference shifts data centers from storage-heavy to real-time processing, greatly increasing energy demand.
- Location of processing versus storage adds complexity for access, cost and environmental impact.