Breakthrough in GenAI fraud and plagiarism: Google's watermarks

Plus: Why watermarks are important yet controversial

Hi AI ethics enthusiasts,

Huge news in the world of AI and AI ethics: Google now watermarks their AI-generated text, making it possible to detect when text was generated by them. They are the first major company to implement this capability.

Today, I'll explain why this watermarking breakthrough matters, how it works, and discuss its effectiveness. Spoiler: I commend Google’s watermarks, but I remain deeply concerned about GenAI fraud and plagiarism.

For dessert, an AI-generated take on this post!

The importance and controversy around GenAI watermarks

AI-generated text creates huge risks of fraud. For example, schools, universities, courtrooms, employers, and many others evaluate peoples’ performance based on texts they write, such as student papers and job candidates’ home assignments. The fact that we can no longer tell (or prove) when a text has been fully written or ideated by the individual who claims to have written it is an enormous pain point.

Last year, there was a surge of tools t…