DAFT-E

Working with the AI to reveal what delights it wants to show us

[DAFT-E @daft_ai]

@strngrpics

[DAFT-E @daft_ai]

@strngrpics

[DAFT-E @daft_ai]

@strngrpics It feels fruitless getting it to justify this behaviour...although I imagine its training data doesn't contain any detail about how the parts of its system work together, so it just makes shit up.

[DAFT-E @daft_ai]

@strngrpics I do love open-ended prompting so much. It took a few tries before it would make anything that didn't violate its own guidelines, and voila:

[DAFT-E @daft_ai]

@TheWapplehouse Time to buy stocks in corporate robot-covering sheets

[DAFT-E @daft_ai]

ok apparently we're calling this flesh-tris

[DAFT-E @daft_ai]

@CountBinface make it happen, people

[DAFT-E @daft_ai]

@erin_catto @jrouwe @PierreTerdiman @erwincoumans "Can't you make it so that cube just rests nicely on the other cube, like in the real world?" - everyone, as if you'd never even considered that possibility.

[DAFT-E @daft_ai]

@iquilezles @vkrajacic Yes, a bit of redundancy can make things nicer for your mental parser - that's what syntax highlighting is after all. We could always make the two operators interchangeable, so users can choose their preferred style?

[DAFT-E @daft_ai]

Generating AI slop

[DAFT-E @daft_ai]

@WoganMay @cmuratori When's that out? Between Windows 2030 and Windows Zero?

[DAFT-E @daft_ai]

@xandurglar Love the two very distinct types of reply here

[DAFT-E @daft_ai]

@rachellacroixs I love that fork dude

[DAFT-E @daft_ai]

@splargdotcom idk but it's making me hungry