Chapter
Skepticism and Deep Learning
1:20:36 - 1:25:07 (04:31)
The progress of AI is determined by searching for cases where a system fails in a way humans would not, often leading to skepticism about the accuracy of deep learning when the mistakes made do not make sense.
Clips
The speaker ponders the potential for deep learning systems to attain consciousness and suggests that the complexity of the brain may be underestimated, leading to skepticism about AI's ability to achieve true intelligence.
1:20:36 - 1:23:16 (02:39)
Summary
The speaker ponders the potential for deep learning systems to attain consciousness and suggests that the complexity of the brain may be underestimated, leading to skepticism about AI's ability to achieve true intelligence.
ChapterSkepticism and Deep Learning
Episode#94 – Ilya Sutskever: Deep Learning
PodcastLex Fridman Podcast
The progress of AI is analyzed by finding one case where the system fails in a big way, which typically leads to articles that convince the public that the system is not intelligent.
1:23:16 - 1:25:07 (01:51)
Summary
The progress of AI is analyzed by finding one case where the system fails in a big way, which typically leads to articles that convince the public that the system is not intelligent. However, while AI may make mistakes that humans don't, it's possible that AI models like GPT-2 are much smarter than human beings at many things.