Last Week in AI GPT-3, Limits of Deep Learning, Deepfakes in the Real World
Jul 26, 2020
PhDs discuss the ethical challenges of AI in healthcare, highlighting the need for transparency and trust. They examine the limitations of deep learning and ongoing innovations, questioning the notion that the technology has reached its peak. A German court ruling against Tesla sheds light on misinformation in autonomous vehicles, weaving in the dangers of deepfakes. The conversation ventures into the revolutionary effects of GPT-3, tackling the complexities of reliability in AI-generated content and its impact on programming languages.
AI Snips
Chapters
Transcript
Episode notes
Explainability and Over-Reliance
- AI explainability is crucial for trust, but current systems lack transparency.
- Over-reliance on AI decision-making tools by doctors is a concern.
AI Regulation and Disclosure
- Consider transparency and disclosure when shaping AI regulations.
- Doctors also withhold information, but AI lacks the ethical training of medical professionals.
Computational Limits of Deep Learning
- Deep learning progress relies on computation, but newer methods are computationally expensive.
- MIT researchers found that achieving incremental improvements now requires exponentially more compute.