AI is not safe now

Steve Eisman · 0m · transcribed 21d ago

Section Insights

# 0:00

The Value of AI in Various Applications

What are the positive aspects of AI in different fields?

AI has beneficial applications in medical problems, GPS, recommendation engines, and chatbots, all of which are legal and widely accepted.

AI is appreciated for its specialized applications.
There are many legal and useful forms of AI.
Public sentiment towards AI is generally positive when it serves specific purposes.

# 0:09

Concerns About a Specific AI Product

What concerns exist regarding certain AI products?

There is a specific AI product that is considered unsafe, raising concerns about its legality and safety.

Not all AI products are deemed safe.
There are legal implications for AI products that cause harm.
The discussion highlights the need for regulation in AI.

# 0:19

Regulatory Approaches to Unsafe AI Products

How can laws be crafted to address unsafe AI products?

Laws could be designed to temporarily suspend products that have been shown to cause harmful incidents until they are deemed safe.

Regulatory frameworks can be adapted to address AI safety.
Temporary suspensions of harmful products are a viable legal approach.
Historical precedents exist for removing unsafe products from the market.

# 0:28

Historical Precedents for Product Regulation

What historical examples support the regulation of unsafe products?

The FTC has previously removed products from the market, such as a decongestant that caused strokes, illustrating the importance of consumer safety.

Historical examples show the effectiveness of product regulation.
Consumer safety has precedence over product availability.
Regulatory actions can be based on health risks associated with products.

# 0:38

Personal Impact of Product Regulation

What is the personal impact of removing unsafe products?

The removal of a previously effective decongestant due to safety concerns highlights the personal consequences of regulatory decisions.

Regulatory decisions can affect individual experiences and choices.
Safety concerns can lead to the loss of beneficial products.
Personal anecdotes illustrate the broader implications of product regulation.

Transcript

0:00 You could say, "Look, we love AlphaFold. We love the kind of special purpose AI for medical problems. That's still good. We love AI for GPS. We love AI for recommendation engines. We even love chatbots for having conversations. All that stuff is legal. There's lots of AI that's legal. But there's this one product" and think about it, it's a product and it's not safe now. You could craft the law in different ways. You could say any, you know, product that has been shown to cause these incidents will be temporarily suspended until such time. Like it's not a crazy thing to do and it's it wouldn't be the first time in history that somebody said you like like the FTC took a drug off the market that I used to like called Tavist, which was the only decongestant that ever worked for me. But they discovered that it caused strokes in young women. So they just took it off the market.

0:45 >> Right.

Summary

The speaker discusses the distinction between beneficial AI applications and those that pose safety risks, advocating for regulatory measures to address unsafe AI products. They draw parallels with historical precedents where harmful products were removed from the market to protect public safety.

- Emphasizes the value of AI in various fields like medicine, navigation, and chatbots.
- Highlights the existence of AI products that are currently unsafe.
- Suggests that laws can be crafted to temporarily suspend unsafe AI products.
- Compares the regulation of AI to past actions taken by the FTC against harmful drugs.
- Advocates for proactive measures to ensure public safety in AI development and deployment.