A "human in the loop" isn't just necessary to manage AI's technical failures or in cases of human responsibility. This paper explores the case where AI provides the "correct" answer to the prompt, but the human realizes important factors that weren't even in the prompt. I interpret this as "rubber ducking," but from the POV of automation.
Thank you, this is the deepest comment in the thread (and lol at all the self-demonstrating sentences).
I'm annoyed that so many people limit their analysis to shallow things like em dashes and "delve," and don't seem to have the critical thinking necessary to analyze textual construction, let alone actual meaning. Instead it's all cutesy remarks about how Emily Dickinson must have been an AI.
I knew Om when I was an ankle-biter writing for a Gawker blog. He was gracious and generous. Felt like everybody's favorite uncle. I hope I can be as kind to others as he was to me.
That reminds me of this post on how Craigslist's business is being eaten away, and why it's taking so long (because of the critical mass needed to compete against its network-effect benefit): http://www.quora.com/Why-hasnt-another-product-disrupted-and...
But if you're coming up with zero ideas, adding three a day is an improvement. What you call "forcing yourself," I call every writers' room and recording artist contract.
I didn't write much good material until Valleywag made me write twelve blog posts a day. Only three or four were really good -- but that's three or four more good posts a day than I was turning out before.