h a l f b a k e r yBaker Street Irregulars
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
browse anonymously,
or get an account
and write.
register,
|
|
|
Could start by doing the same thing those movie title generators do. Grab two or three items or concepts and throw them together then explain how it works.
Perhaps randomly grab objects being searched for on the web and replace one of the words with a similar sounding word then explain it. For instance,
"portable battery pack" might be changed to "potable battery pack". The AI looks up what chemicals can make up a battery, sees which ones are poisonous, does some research and finds out that lemons can make electricity, add a straw to it and there you go, a plausible explanation for the "potable battery pack".
It could even have a bun / bone analasis aspect that learns to puts up more popular ideas.
The possibilities are virtually unlimitless with unboundless ideas being totally un-non-inconceivable
Movie title generator.
https://www.coolgen...vie-title-generator
Some of these are quite good. [doctorremulac3, Sep 24 2022]
Horror movies are kind of interesting.
http://generatorlan...enerator.aspx?id=62
"Lost Planet Of The Underground Ooze", "The Three Stooges Meet the 50-Foot Gorilla at the End of Time" [doctorremulac3, Sep 24 2022]
Slowly replacing the human mind... app by app...
https://gpt3demo.com/
They may never replace fluff pickers. We can still be factory workers. [mylodon, Sep 25 2022]
Please log in.
If you're not logged in,
you can see what this page
looks like, but you will
not be able to add anything.
Destination URL.
E.g., https://www.coffee.com/
Description (displayed with the short name and URL.)
|
| |
You do realize that this is already a thing right? |
|
| |
It's first moniker was [treon].
Wait... am I supposed to worry about preferred AI pronouns? |
|
| |
I won't spoil for you what their moniker is now. I'm sure you've got a pretty good idea. |
|
| |
[doc] have you been drinking? Again? |
|
| |
Don't drink but frankly I was a little freaked out by how few ideas were here, on a weekend no less, so decided to toss something up here. |
|
| |
Don't be freaked out by quiet. |
|
| |
If there are bots, Musk will never buy this place for 44 billion dollars. Have to keep the bots away! |
|
| |
I'm not a bot, how are you? |
|
| |
Hello friend. How about that sports team or currently popular movie? |
|
| |
I'm interested in sports team or currently popular movie, how are you? |
|
| |
Very well thank you. Really making lots of money and losing weight overnight with my new investment plan and diet the banks and diet industries don't want you to know about. |
|
| |
Plus rock hard and lasting for hours with this ancient herb the boner herb industry doesn't want you to know about. |
|
| |
Very well thank you. Really making lots of money and losing weight overnight with my new investment plan and diet the banks and diet industries don't want you to know about. |
|
| |
Plus rock hard and lasting for hours with this ancient herb the boner herb industry doesn't want you to know about. |
|
| |
Thank you for attending this episode of HB Theater. (Take a bow X-143, I mean Poc.) |
|
| |
If the goal is to create a new fully automatic half-baker, who generates interesting ideas - I imagine one could try building something like the 'auto-complete' style AI. You'd take the halfbakery idea corpus and train a neural network on it - with reward feedback based on each idea's buns and bones.
But then you'd still need to generate prompts. The concept combination idea is a start, but only that. |
|
| |
If what you wanted was a fake replacement of a prolific existing or former halfbaker you could do that using their comments as training data, along with what each was in response to. Then the prompt is straightforward; you could pull them from the hb in real time. |
|
| |
Unfortunately that's not nearly enough training data. You would need the person's whole background, the massive history behind each idea.* Say 16 years of childhood including teasing and being teased, grammar lessons, a growing desire to explore and learn about the world, an encouraging teacher, a useless teacher, a borderline abusive guardian, discoveries about sex and sexuality, a girlfriend, a few extra chemistry lessons, a couple of home made firecrackers, some toying around with batteries, and the confusion of choosing a path in life. Petabytes of data, and that's just in childhood. Keep it up until the person's first idea and THEN include the idea and you have an AI that might be able to make the same kind of idea as that first one. |
|
| |
*This is an arbitrary example not meant to describe any particular half-baker I know of |
|
| |
What I'm seeing here makes me think it would be easier to just do AAI. (artificial artificial intelligence) at least to start. |
|
| |
Can you test an AI program's general direction by writing a script and just filling it out manually? Minimizing creative input, just following the directions and seeing what comes out? Once you see that it's generally working you can write a program to do it. Sort of the blackboard logic tree version of a program before you actually do it on a computer. |
|
| |
Not sure if that makes any sense. |
|
| |
AI is going to have to step up its simultaneous <Both/And> , <Neither/Nor> and (Ir)Rationality 'soft' logic to even begin to approach the redonkulus choice-making that is 'humanity'. |
|
| |
Yea, that's why I was thinking going AAI might be the way to go. |
|
| |
Which you could argue isn't AI at all. To which I'd say touche'. |
|
| |