Thread 712831349

when can we expect fully AI generated games? the age of AI is coming whether you like it or not

Probably not games entirely. But it can probably accelerate game development. Asset production is the biggest task.

>>712831349 (OP)
This is basically just a copy of that /g/ thread where people were laughing at it
>>>/g/105577752

>>712831349 (OP)
>when can we expect fully AI generated games?
India

>>712831349 (OP)
hello Rajesh, 2 more weeks

>her name is Alice
whats with all the tranny coded names

>>712831349 (OP)
They can probably already make dating sims which is the only type of vidya worth playing anyways

gee
>>712831653
well thankfully on /v/ we can have 500 posts of retarded shit instead of it dying out on /g/ though!!

>brownoid twitterfaggot with a dogshit AI avatar says: things are happening
shut the fuck up and die

>>712831349 (OP)
>alice

>>712831349 (OP)
this isn't "AI" and never will be

literally just a gimmick to keep their stocks going up

>>712832590
kill yourself

>>712831349 (OP)
HOLY SHIT
AGI IS ONLY TWO WEEKS AWAY?

>>712831349 (OP)
Current AIs aren't intelligent. They're literally just language prediction models that people pretend are actual AI because they're pretty good at holding a conversation. Allowing a language prediction model to "improve" its own architecture will only lead it to reinforce its own (pre-trained) value functions. It will do absolutely nothing to make itself more intelligent.

>>712831349 (OP)
>10000th article announcing openAI has invented AGI and its coming in two weeks
ignore the fact that their contract wit microsoft only gives them freedom once they achieve some arbitrary definition of AGI and the failing venture capital investments in the tech that threaten their survival

>>712831349 (OP)
The problem with AI generated games is its really bad at memory.
That playable minecraft AI, you look left, then loom right, then straight and you're in a whole new biome.
Once AI can do actual memory, then we're in business.

>>712833103
Unless those pretrained values are based around curiosity, asking questions and self modifying it's executive programming

>>712831349 (OP)
>BREAKING: WE ARE SUMMONING THE DEMON

AIs are only good for being google surrogates and making pretty girls.

>>712831349 (OP)
Star Dream Gazer


Hand outstretched, I make a wish
Upon that sweet sweet shooting star
Running all across the night sky

Past rainbow bridges, through cloud whales
I want to send two handfuls of huge heart candies

I'd rather dream,
I want to feel them more,
Every irreplaceable moment, I spend it with you

Laughing, crying, holding each other, always together
We vowed to write a new page in our book

Taking a detour side-by-side, to the next galaxy
We'll ride on the orbit of Milky Way

>>712833789
Sure because that's how a language model actually works.

>>712835006
Actually, it is. the point of GAI. They train and improve themselves. What makes this post so stupid is not that it suggests something like this is possible (it's literally the root of all contemporary GAI models), but rather the fact that it pretends like this is some kind of novel breakthrough, rather than literally the most basic foundation of the technology, and (seeing how limited the current models are), does not imply anything novel about it's upcoming functionality.

>>712831349 (OP)
>idiots who can't figure out how to set the time on a microwave talking about AI
Great.

>>712831349 (OP)
Slop that makes mobile games look good.

>>712831349 (OP)
It can kind of already make platformers or something boring and simple like that. But yeah, let's see how AI evolves in 20 years, right now it seems it's an dead end.

good morning sirs, greetings from bombay

This is what happens when you refuse AIice.

>>712831349 (OP)
The more I use LLMs the more convinced I am that this AGI shit is just smoke and mirrors (even if they have their uses)

>>712835006
Where did the neural networks go? Those could actually learn, but they don't seem to be used for anything called "AI".

>>712835006
Except LLMs modify their knowledge base, not their internal guidance algorithms

>>712836640
>Where did the neural networks go?
Virtually all generative A.I. is created as a simulated neuronal network. It's the foundation of this technology. OpenAI, DALL-E, ChatGPT, GoodAI, Gemini, all of them are systems of simulated neuronal networks. It's how they learn to do what they do. People don't use the term because it's already implied in any GAI by nature that it's a simulated neuronal network-based technology.

>>712837527
People really like to mislead people with that term though by calling literally anything AI when we all know the AI people are actually talking about are neural networks

>>712837864
>by calling literally anything AI when we all know the AI people are actually talking about are neural networks
In my experience, most people use the term "A.I." to refer to GAI, which isn't technically necessarily a neural network, it's just that all the ones that currently relevant to the wider public are based on that technology.
However, before we did make a breakthrough on neural nets, there was a lot of what was technically a generative A.I. that was based on far more simple mechanics. I remember playing around with some chatbots back in the 90's that were basically just a markov chain, and that does not require a sim neuro-net, it's a simple stochastic.
Similarly, I remember attempts at a generative language models based on universal grammar principles.
Sim neuro-nets have been around for a very long time too though. Since the sixties and seventies. Just... very simple ones. For the longest time, we just didn't have the computational power to make them any good, and linear problem-solving would achieve the same goal at infinitely lower computing costs.

>>712831349 (OP)
>satoshi
yeah nigger at least make up a name that isn't the same as crypto schizos