567: Kai-Fu Lee | Ten Visions for Our Future with AI

Genetic Data

Privacy concerns surrounding genetic data are significant, as Kai-Fu Lee explains. Genetic sequencing is highly sensitive and cannot be anonymized, making it vulnerable to misuse by unethical entities or nation-states 1. Lee suggests privacy computing and federated learning as potential solutions to protect this data while still allowing AI to train on it 1. Jordan Harbinger expresses skepticism about the security of such sensitive information, especially given the potential for hacking and misuse 1.

We need to boost security everywhere. I mean, people might think, "Hey, if I just stored my genetic sequencing on my phone, then I feel safe." But actually that's the easiest thing to hack into, right? Even worse than hospital IT.

--- Kai-Fu Lee

The discussion also touches on the vast amounts of data required for AI development, emphasizing the need for relevant and secure data sources 2.

   

Bias Mitigation

Addressing biases in AI is crucial for fair and effective systems. Kai-Fu Lee compares training AI to raising a teenager, highlighting the need for balanced supervision to filter out harmful content without overly restricting data 3. He emphasizes the importance of training AI on diverse and balanced datasets to avoid inherent biases 4. Jordan Harbinger raises concerns about cultural mismatches in AI trained on data from specific regions, which Lee acknowledges as a challenge that requires global data collection and balanced training 4.

A good example is TikTok, right? That is a product that is global. Actually their version for China is very different from the version of the US, not only is the training data different, but usage habits are different.

--- Kai-Fu Lee

Lee also suggests that AI engineers must be educated on their responsibility to ensure fairness and that tools should be developed to automatically detect and correct biases in AI systems 4.