GPT-2 Text Generator With Api...

This feature request has been completed. Install the latest ARC Early Access to experience the newest features and bug fixes.

Get ARC Early Access

Hey I found is not really a feature request, but I thought it might be interesting for you guys to try?
Its also an attempt to not open a thread for a question, but using this way for putting some content to the forum!:)

They are having a nice I figured it could be nice for building some oracle, or poem device.
Since the answers are sometimes a bit of, but really funny!!


Upgrade to ARC Pro

Experience early access to the latest features and updates. You'll have everything that is needed to unleash your robot's potential.

Bahahaha that is hilarious!!!
I mean - it does sort of make sense. There's logic behind it but, wow... just wow!
#6   — Edited
User-inserted image

*edit* just realized the ai dropped the f** word and edited it out:D
xD There is some real wisdom in that one too!!!:D
Sunspring. A couple filmmakers fed an AI Sci-Fi scripts then had it create a new one. ....its so so so.....indescribable must watch it! This was 2016, curious with advancement in AI what a new attempt would produce something a bit coherent ?! 

#9   — Edited
Well - ai in the sense we’re discussing can’t really be improved much because the neural net is merely associating existing sentences that it’s been fed. General intelligence is (agi) is the advancement we’re all hoping for. Because agi, much like Perry’s interest, is about understanding properties of things. Meaning, ai says this word is near this word x percent of the time. Where agi says this this property of this thing is associated with this thing’s property because of this other thing and can be used for these things etc etc

another example is a glass on a table. Agi knows the table and glass are separate items and it also knows basic usage information of the items. The AI would not know the items or that they’re even separate. The ai may say a table with a cup on it, only because that’s what it knows that image to be. It doesn’t know they’re separate items. 

so when you break down this scenario of conversing with a computer, ai isn’t really that useful outside of a predetermined goal. However, for general conversation that has no goal and can kinda go anywhere like two humans talking, you need agi

that service we’re playing with appears to have a database of existing complete sentences that it’s retrieved from scouring the internet and RSS feeds (presumably). So it can relate similar words of similar articles and extract sentences to match them together. It’s actually quite basic but real funny

albeit, the funny part is only due to false expectation that it should work. We read into the sentences and TRY to relate the randomness to the topic. Like a Buddhist monk talking in riddles...

remember, ai is merely a huge database of related things it’s been taught. So it can appear random, but it’s not - there’s association. The more data and complex the network is, it’s more difficult to predict.
Yeah when AI can truly understand the context of a sentence (or an image) and can relate it to a subject/s, then we are talking about an understanding of what is being said/written. Its very interesting to be present and watch the birth of AI. Because I feel like we are only at the single cell stage! Lets hope for exponential growth.
#11   — Edited
That’s not in the definition of ai. Again you’re talking about agi. That’s like saying it’ll be great when cars become trucks so they can carry more. Cars are cars and trucks are trucks.:) Ai is ai and agi is agi 

ai is already a thing. It’s a statistic based network of associated items. Ai has its use, and mostly in finance these days...

agi is the thing you’re excited to see.
...........Got it!
A link so nice, it was pasted twice
This is a nice addon to the discussion my sense GPT-3 is already a step forward to Artificial General Intelligence, since it is very versatile. At least in pretending to be intelligent and sentinent! 
Lets see how GPT-4 will Lex Fridman says, training data is the bottleneck!!:D
#19   — Edited
We are in the early days of GTP-3, 4 - the future may hold better the key to unlock this conundrum.

getting old - I been in the right place, but it must have been the wrong time  :-)



import requests, json

def generate_text(text):
endpoint = ""
# refer to to customize your data request
data = json.dumps({
"prompt": {
"text": "{}".format(text)
"length": 100
headers = {"Authorization": "Bearer ************************************"}
response =, data=data, headers=headers).json()
return (text + ' ' + response["data"]["text"])

print(generate_text("hello how are you?"))
Just wanted to share this, to save anyone wanting to call the api some time!!:)
It says this feature request is "Completed". If so where would I find the skill? I have looked through the AI skills and it is not there. Is it on hold? What are the plans if any to add any of the GPT (2? 3? 6? x?) code as an ARC skill?

Oh my God, too funny but hey it works!.
" Gypsy priests may not be permitted to use holy water to bless their hands after mass as they are said to have a laxative effect on nuns in that church

Temple Chaiyaphum temple halts religious ceremony after a..."

I really should get my robot to speak that in Cylon voice!!xD