AmiSight 10/5: The Real AI Fear May Be Liability
We hear a lot about the need to regulate AI, but Andy Kessler makes an interesting argument in The Wall Street Journal: The force that may ultimately keep AI companies in check isn't regulation. It's liability.
Kessler compares AI to self-driving cars. With Level 2 autonomous driving, the driver is still expected to pay attention and take responsibility. Once you move toward Level 3, the car takes over more of the decision-making, and suddenly the question of who is responsible when something goes wrong becomes much more complicated.
The same thing is beginning to happen with AI. There is a big difference between asking ChatGPT to help draft an email and giving an AI agent the ability to send emails, access systems, execute code, make purchases or take other actions on our behalf. The more autonomy we give AI, the harder it becomes to simply say that the human using it is responsible for everything it does.
Kessler's argument is that the threat of lawsuits may ultimately create a stronger incentive for AI companies to build safeguards than pages and pages of government regulations. If a company knows it could be financially responsible when its AI causes harm, safety suddenly becomes more than an ethical discussion. It becomes a business imperative.
For entrepreneurs, I think there is a broader lesson here. As AI moves from being something that advises us to something that acts for us, we need to start asking a different question. It isn't just, What can AI do for my business?
It's also: Who owns the consequences when it gets something wrong?
That may turn out to be one of the most important questions of the AI era.
Comments