4 ms·
I tried recreating one of my favorite strips, just to see what would result. This was the result: https://create.shortbread.ai/viewer/41e5642c-86ee-421e-b985-b
by iamwil 3y ago
I tried recreating one of my favorite strips, just to see what would result. This was the result:
https://create.shortbread.ai/viewer/41e5642c-86ee-421e-b985-b39011d5ea18 https://create.shortbread.ai/viewer/41e5642c-86ee-421e-b985-...
The original (SFW, despite the warning):
https://www.oglaf.com/trapmaster/ https://www.oglaf.com/trapmaster/
The writing obviously will need work, but I can see this being much better in six years time. The fact we can do this at all is pretty neat. And honestly, a lot of people have a hard time going from zero to one. It's much easier for most people to fiddle and edit something, than to stare at a blank page.
I suspect this kind of workflow will permeate all kinds of work, where the generative AI gives us a starting point, and we go from here. The question then, is whether starting points matter, because it's still a far gap to get to the Oglaf strip given the current starting point. Would some people fare better by random? Or we can make AIs that have enough context to know what's fresh for people?
- Fengjiao 3y agoI think the baseline can def be improved. But I suspect the gap between 60% and 99% would be so wide that it would still take a significant amount of time for the human creator