chisel

It’s long been proven that LLMs can produce exact copies of their training data. Not every piece of training data, of course, but enough where spitting out exact replicas of proprietary code is a legitimate concern. Not to mention that copies don’t need to be exact in order to be a problem.

You could have said the exact same thing about Digg or MySpace 15 years ago.

Reddit is difficult to replace, but not nearly as difficult as you imply. Serving a lot of web traffic with low latency is a well documented and solved problem. And a competitor does not go from zero to 100 million users over night, they have plenty of time to scale. Countless companies have accomplished this.

Reddit is, by and large, an extremely basic concept. It’s a link sharing site with comment trees. Undergrads could spin up a similar website in a week, albeit less scalable. But again, there is plenty of time to figure out the scaling issue.