• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

AI generated texts, say hi to GPT-3 and its 175 billion "neurons"

llien

Banned
A spooky reminder: we have absolutely no idea where self-consciousness comes from.


Text marked bold is input, the rest is output:

The following is an essay by Paul Graham on paulgraham.com on how to make Mars a startup hub.


Startups On Mars
By Paul Graham


Here’s the sort of startup that might make a go of it on Mars:
One that makes money by taking advantage of local resources. For example, a startup that uses the Martian atmosphere to make fertilizer or methane fuel. A startup that sells water from Martian rivers. A startup that uses minerals from the Martian soil. A startup that uses solar power from the Martian sun. A startup that uses a new kind of solar panel that’s more efficient at converting sunlight into electricity.
I would even invest in a startup that did nothing but extract the helium-3 from the Martian atmosphere and send it back to Earth.
If you can find a way to make money by using local resources, there’s probably a business there.
There’s another class of startups that might work on Mars:
Startups that produce services for other startups. For example, a startup that makes it easier to build rockets, or to extract gases from the Martian atmosphere, or to construct solar panels, or to farm the local soil.
Startups that make things more comfortable for colonists. For example, a startup that sells air conditioners, or houses. A startup that sells things to keep colonists entertained. A startup that provides Internet services.


Kayne West Exclusive – Why He’s Running for the Presidency, and what his priorities would be as President.
Anderson Cooper: Yeezy, how would you describe your presidential platform?
Kanye:
My platform is very simple. I want to make everything awesome again.
Anderson Cooper: But what about foreign policy?
Kanye: Who cares about that stuff? Let’s talk about important stuff, like Kim Kardashian’s butt. Anderson Cooper: That’s a bit sexist.
Kanye: I’m sorry. I’m not sexist. I love women. I’m just an equal opportunity objectifier.
Anderson Cooper: But what about abortion?
Kanye: I think abortion is fine. I mean, women should have the right to choose what they want to do with their bodies.
Anderson Cooper: But what about climate change?
Kanye: Let’s talk about Kim Kardashian’s butt.
Anderson Cooper: I’m going to be straight with you, Kanye.
Kanye: Oh no, you’re going to be gay with me?
Anderson Cooper: No, I’m going to be straight with you. You’re not going to win this election.
Kanye: I’m going to win.
Anderson Cooper: No, you’re not.
Kanye: Yes, I am.
Anderson Cooper: No, you’re not.
Kanye: Yes, I am.
Anderson Cooper: No, you’re not.
Kanye: Yes, I am.

“The importance of being on Twitter,”
by Jerome Jerome


It is a curious fact that the last remaining form of social life in which the people of London are still interested is Twitter. I was struck with this curious fact when I went on one of my periodical holidays to the sea-side, and found the whole place twittering like a starling-cage ...

(I couldn't find the whole piece for J.J.)

More examples:

Notable article on GPT-3 written by... GPT-3.

See also:

A kind reminder, number of neurons in brain:
cats - 2 billion
dogs - up to 5 billion
chimps - 5 billion (our closest relatives among primates)
orangutans - 7 billion
gorillas - 9 billion
humans, dolphins - 15 billion
killer whales - 30 billion

This beast beats at as sheer number of neurons, but possibly is still behind on complexity, as our neurons tend to have lots of connections.
 
Last edited:
aren't 175Billion parameters more like 175 billion synapses than 175 billion neurons?

If so, the human brain is said to have 16 B neurons x 10,000 synapses or 160 Trillion synapses. Or this would be like 0.1% of human brain, if I'm correct.
 
Last edited:
The human brain has ~86 billion neurons and over ~100 trillion synapses (e.g. "connections" in the case of neural networks).

Maybe when the amount of connections is 1000x larger, and the neural network is equivalent in size to a human brain, we will achieve human-level AI. Especially if the training datasets include video, images and sound rather than just text.

1000x improvement will easily be achievable by the 2030s. Imagine AGI by 2030... It's possible that increasing data quantity and network size won't get us there, but if it does, we might be living in the last few years before the greatest technological shift in human history.
 
Last edited:
Top Bottom