[usaar333]

I find it weird to compare GPT-4 to not expert humans (which this did - they seemed to have just sampled pitch decks for funded companies). GPT-4 generally will beat non-experts.

What you want to see is a comparison to say pitch deck consultants - it's also more useful to evaluate whether AI or a human should help you build one.

This does suggest at minimum though that if you are a founder that GPT-4 will likely provide positive value in helping you relative to no one.

[esjeon]

> GPT-4 generally will beat non-experts.

Just a nitpicking, but I would say GPT can easily beat untrained people. Clearly, GPT has already learned from all the good examples available on the internet, and it's nothing weird about that trained ones mostly outperform untrained ones.

[ftxbro]

yeah nothing weird about these LLMs at all

[golergka]

If a company was funded, doesn't it mean that pitch deck was successful?

[usaar333]

Only on a binary scale. There's even among actually funded companies better or worse pitch decks.

You can often just get funding by having a good team matched to a VC that knows your industry. Pitch deck can be crap and other VCs might hard reject.

Another potential methodology flaw is that they don't seem to group decks by VC industry. I definitely altered my deck for different VCs, in ways which could make it a worse deck for the "average VC". GPT-4 might be better at setting context that industry expert VCs don't need - meaning this isn't even comparing like and like.