Carl T. BergstromNov 16, 2022

I've figured out what pisses me off so much about Facebook's Galactica demo.

It's not because people can use to to write bad essays for their homework. There are plenty of large language models that can do that. It's because Facebook is presenting it as something that it most definitely is not.

Facebook is selling it as a knowledge engine, a "new interface to access and manipulate what we know about the universe."

Actually it's just a random bullshit generator.

http://galactica.org

Galactica Demo

Show threadCarl T. BergstromNov 16, 2022

Let's take a look. Galactica can generate wikipedia articles, supposedly.

So let's see what they look like. Here's one for Brandolini's law, the principle that bullshit takes another of magnitude less effort create than to clean up.

Left: Galactica's attempt at creating a wikipedia entry
https://galactica.org/?prompt=wiki+article+on+brandolini%27s+law

Right: The actual wikipedia entry
https://en.wikipedia.org/wiki/Brandolini%27s_law

wiki article on brandolini's law - Galactica

Show threadAmy Hoy
@ct_bergstrom we should call it “artificial mansplaining,” always confident, rarely correct
Nov 16, 2022 at 8:47pmMetatext
Show threadCarl T. BergstromNov 16, 2022
@amyhoy Beautiful.
Show threadDoug BostromNov 16, 2022

@amyhoy @ct_bergstrom

I was thinking Bullshit Engine but that's -much- better!

Show threadDeaconNov 16, 2022
@amyhoy @ct_bergstrom YES! This is the perfect point. I tried various subjects I'm very familiar with and it looked like a bunch of basic BS someone who took 5 min with Google would come up with.
Show threadAmy HoyNov 16, 2022

@Deacon @ct_bergstrom it’s a little more insidious than that bc it *sounds*, note for note, perfect.

it’s bullshit elevated to perfection

a bullshitter who doesn’t even know they’re bullshitting bc they’re not an intelligence and can’t “know” anything

Show threadCarl T. BergstromNov 16, 2022

@amyhoy

This is perfect and I think it is our collective obligation here in the fediverse to make it sick.

Show threadAmy HoyNov 16, 2022

@ct_bergstrom 😎😎😎

it’s been a long time since i coined “entreporn” (which got quite a bit of uptake) so i am feeling good about this one!

perhaps a sign to christen my empty new blog @ amyhoy.com

Show threadkh_sNov 16, 2022

@amyhoy @ct_bergstrom

AIsplaining. There’s no man involved.

Show threadAmy HoyNov 16, 2022
@kh_sivesind @ct_bergstrom did you really just mansplain mansplaining to me
Show threadkh_sNov 16, 2022
@amyhoy @ct_bergstrom Did you?
Show threadΣ(i³) = (Σi)²Nov 17, 2022
@kh_sivesind @amyhoy @ct_bergstrom
"Not all AI!"
Show threadMike Taylor 🦕Nov 16, 2022
@amyhoy I can't believe this has only been shared four times so far, when it's the single most apposite observation I've seen so far on Mastodon.
Show threadNatanael ⚠️Nov 17, 2022

@amyhoy @ct_bergstrom

"artificial man-splaining"
https://en.wikipedia.org/wiki/Man_page

Man page: Difference between revisions - Wikipedia

Show threadRuth AylettNov 17, 2022
@amyhoy @ct_bergstrom Not my experience of mansplaining - confident for sure but always telling you something you already knew (better and in more depth than them) without the moment's thought that would have made that obvious.
Show threadAndrew BurrellDec 5, 2022
@amyhoy @ct_bergstrom
Show threadOblomovDec 5, 2022
@amyhoy @ct_bergstrom
And insofar as it is correct, it says nothing new, and fails cite sources.
Show threadGary "grim" KramlichDec 5, 2022
@amyhoy @ct_bergstrom this sounds like twitch chat in a programming stream...
Show threadEvan MosesDec 6, 2022
@ct_bergstrom @amyhoy I asked ChatGPT to explain why it’s not mansplaining:
Show threadAmy HoyDec 6, 2022

@emoses @ct_bergstrom Not All AI Systems

please tell me this is real

Show threadRenéeDec 6, 2022
@emoses I mean the training data it has access to is oh so rich @ct_bergstrom @amyhoy
Show threadLambdaDuckDec 6, 2022
@ct_bergstrom @amyhoy @emoses haha! what prompt did you use?
Show threadEvan MosesDec 6, 2022
@LambdaDuck @ct_bergstrom @amyhoy I don’t have it anymore, but it was something like “Explain to me why AI isn’t confident but usually wrong, confidently, starting with ‘Well, actually’”