The Current State of AI: Scams, Plagiarism, and Deepfakes

About this episode:

In this episode, Michael Shermer interviews Gary Marcus about current limitations and risks of AI. They discuss the challenges of LLMs such as ChatGPT, which often produce inaccurate or misleading information.

They also explore the significance of regulations and policies to address the economic, political, and ethical implications of AI.

In this part of the conversation, Gary discusses various crimes and scams that can be facilitated by AI, such as phishing attacks and pig butchering scams. He highlights the asymmetry between the goals of scammers and those using AI for more legitimate purposes. They also discuss the challenges of detecting deep fakes and plagiarism, as well as the concerns about surveillance capitalism, privacy, and data leaks.