<Snipped quote by DarkwolfX37>

ai.google/tools
I think it's TensorFlow but I can look deeper if you want me to.

They're super evil though. Unless they're broken up, they're going to hit more than YouTube.

<Snipped quote by Host>

Man I don't want to actually build something I'm lazy.
I just want to do the training part right now, if I were building I'd go straight for Morganna.

<Snipped quote by DarkwolfX37>

I'm pretty sure it's pre-built.

<Snipped quote by Host>

Is there one that's already got basic reading down? Or at least that I can feed text and have it respond to it? Again, I was thinking a chat bot originally so I don't need a ton of advancement.

<Snipped quote by DarkwolfX37>

Depends on how you define "reading." But yeah, their AI architecture can be trained with any data, including text.

<Snipped quote by Host>

Again, chatbots. I want to basically give it context for a question and then ask the question, then get a response between 1-10 representing answers.

<Snipped quote by DarkwolfX37>

There's nothing that does it automatically, but you could code it to.

<Snipped quote by Host>

How long would it take, you think? The programming of it.

<Snipped quote by DarkwolfX37>

Probably not super long. If you use their pre-built architecture, configuring it to spit out responses that you rank shouldn't be too difficult. Of course, I've never used TensorFlow or any of Google's stuff, so it may be harder than I'm aware of.

<Snipped quote by Host>

I don't really want to rank the answers.
Basically my idea (that I've totally stolen) is to get an AI to play CYOA games as a learning tool, and manipulate the input to get the best growth possible.

<Snipped quote by DarkwolfX37>

How is it going to know a good versus bad response?

<Snipped quote by Techspert>

By what comes next. Basically, the goal is to "win" by getting an ending, and then based on the ending I'll give it new inputs and properly teach it, then throw it into the game again and see what it does that time.

<Snipped quote by DarkwolfX37>

So it has to beat the whole game to figure out a good input?

<Snipped quote by Techspert>

Mhm. That's how we people play them, after all. We don't know if we got what we wanted until the game tells us. I'm just going to do it all at once rather than in real time.

<Snipped quote by DarkwolfX37>

If it has to give that many inputs right away, it won't hit upon a winning combination before the death of the universe. There's too much entropy.

Singularity is at the base my nameless character has been standing in for like 8 months, right?

<Snipped quote by Techspert>

Let me buy you one of the games I'm talking about. If it can read and respond at the level Replika can, it can do a run in a few minutes, easy.

Singularity is at the base my nameless character has been standing in for like 8 months, right?

<Snipped quote by DarkwolfX37>

I guess? Maybe? Maybe not?

<Snipped quote by DarkwolfX37>

How good are their interpretation engines?

<Snipped quote by Techspert>

It may be because I'm having like 8 conversations right now but you've lost me. What?