The AI boom: what and why?
summary
Generative AI accelerates after GPUs enable deep learning around 2013 and a visible boom with ChatGPT (Nov 2022); foundations are data, knowledge (logic), and programming. Real deployments include a one-day-ahead bike-station overflow predictor (Bicing one-day overflow prediction) and an automated survey-to-report system that produces full reports an hour after collection. Risks: models hallucinate, leak prompt data, and reproduce biases — e.g. a soap dispenser failing on dark skin and facial recognition failures highlighted by Joy Buolamwini — and regulators classify high-risk systems while banning social credit; users number 400 million weekly users and failures have led to Netherlands 26,000 families falsely accused.
AI-written summary · about
chapters (14)
- 0:00Introduction and speaker background
- 1:17Digital society and AI's role
- 3:03History of artificial intelligence
- 7:05Generative AI and deep learning boom
- 10:56AI in business transformation
- 13:54Explainable AI and human oversight
- 15:18Generative AI limitations and risks
- 17:23AI applications beyond generative models
- 21:17Data quality and representativeness
- 22:41Anonymization and privacy risks
- 24:27AI biases and social inequality
- 30:36AI failures and ethical consequences
- 34:58European AI regulation and ethics
- 38:11Responsible AI adoption
related talks
19mMachine Intelligence With Humans in Mind
42mFrom theory to practice: The rise of Generative AI
17mGenerative AI: Keeping it real
19mLet's Talk GenAI
29mThe Rise of AI: A Reality Check on Generative AI
20mHuman-Centered AI for Sustainable Development
01h 14mLatest trends in AI
15mPractical Evolution of Data Science with GenAI