writing

seeing more people talking about using llms to produce final written content (not to do research).

for me, this has been v unsatisfactory in writing anything meant to develop/communicate a new concept. 

is it just me?

not-knowing ✹ 🍕-fan ✹ productive discomfort + idk ✹ Marseille / Singapore / Auvergne ✹ The Uncertainty Mindset (2020) ✹ now: vaughntan.org/now-ish

This is so sad
I think LMs can do 90% of creativity in theory, just not the top 10% which requires some ineffable genius

But the relative ease at using an LLM is that the (unique) type of genius an LLM produces will become more common, but more than that, so will the mediocrity it produces, which saddens me most

Be specific.     

leonasskau.co.uk //
Hosting /strategy, /leo, /geopolitics // built Culture3.com // online: leosn.xyz
d33m:mun

i genuinely wonder if the 10% is the actually creative part and the rest is the enabling work

and how much of learning how to do that 10% requires doing the remaining 90% that is just drudgery

I think our standards should not be so high

There is much value in creativity that isn’t at the level of Bach or Steve Jobs

But a very interesting point about how can you BECOME someone who does that too 10%, if you can’t learn your craft
@cl is working on an essay on this topic

I think (A) is simple for AI, and it can easily remix old ideas, like much creativity does, in ways that humans haven’t/might never

But (B) feels impossible for a neural net until social norms shift. But how much creativity should require meaning?
Is it too nostalgic to imagine past centuries were full of meaning?

i'm truly not talking about that level of innov

being able to decide that something new is _good_ requires A) knowing what's new (maybe AIs can be ok at this, depending on their training set), and B) the 10% (or 1%) that can create meaning out of nothing

this is relevant even for even the most trivial creation act