Module 4 / Lesson 11.2.2
Externalities Problems from AI
The externalities problem in AI arises when the private incentives for the people driving AI development fail to account for the broader social costs or risks of their work. Companies and governments often reap the immediate benefits of AI innovation (profits, influence, efficiency), while the long-term harms such as misinformation, labor displacement, or even catastrophic risks are borne by society at large.
This creates a classic market failure: individual actors (companies and inventors working on AI) underinvest in safety, transparency, or long-term alignment because their costs are borne by someone else. Understanding externalities is crucial to designing governance structures that realign incentives with collective well-being.
Learning: Watch the embedded video by Acemoglu on AI and Social Harms.
Reflection: As you watch, focus on
- Note down at least two examples of externalities Acemoglu highlights. How do these relate to the idea that firms underinvest in socially beneficial outcomes?
Learning: Acemoglu on AI and Social Harms
MIT economist Daron Acemoglu has written extensively about how AI may amplify inequality and create widespread negative externalities if left unchecked. In the following lecture, Acemoglu discusses how the benefits of AI are often privatized, while the harms, such as job loss and weakened democracy, are socialized.