Today, one of my final papers came back. I got a grade that I am really happy about. The only issue my professor had was that I did not italicize the title of the newspaper that I was writing about throughout my paper. Yet I knew before handing it in that newspaper titles are supposed to be italicized. Leaving the title improperly formatted was an intentional signal to him that I, not ChatGPT,…
I have notifications turned on for both the New York Times and the Wall Street Journal. Every day, it feels like I receive at least five "BREAKING NEWS" notifications. Yet often these are just regular stories that aren't that interesting. I feel like ten years ago, Breaking News actually meant, well, Breaking News. Something urgent, ground-breaking, happening in real time. The 24-hour news cycle…
In the early days of LLMs, the scaling hypothesis was not initially intuitive. Scaling, in very simple AI terms, is the idea that throwing more data and compute into a model will make it better. Surely, super-intelligence would be achieved through complex architecture, not just... more? However, the huge jump from GPT-2 to GPT-3 was not due to a fundamentally different design; the model had simply…