Paused for VP recharge + some progress on project testing.

Reminder: this account is only for testing AI and bots, for a bigger and more serious project later on. You can upvote it, downvote it, or ignore it. I only do 3% votes, with no HP, purely for testing and with zero influence.

Goal:
Automate intelligent curation using AI as a tool.

This account will disappear later and be replaced by a clean, polished one.

I added a public blacklist via @hivewatchers to avoid voting on certain accounts. It seems to be working: if it detects something, it puts the account in the “not eligible” box.

One thing I quickly learned: an automated message after every vote attracts downvotes very fast. That issue will be fixed later so it doesn’t bother anyone.

Next step (not integrated yet, but already programmed separately):
How to sort posts by theme.

Not by tags, because that would be easy.

But by what the post is actually saying.

I tested keyword detection and it works, but in reality it still just sorts by keywords. Example: after manually rereading a post, it contained the words:
Art, crypto, BTC.

So it got sent to these categories:
#art #ccc #crypto #trading

Even though the post only talks about the price of BTC and future analysis. Logically, only #crypto and #trading should have been used, not #art or #ccc. That’s the flaw of keyword-based sorting. I want it to actually read and understand the text to sort by theme.

I’m looking for a solution using multiple AIs:

one AI that codes,
one AI that tests the code on a private server and detects all errors,
a third AI where I send the errors so it tries to fix them until it works.

Then I report everything back to the first AI and restart the loop. I’m testing a lot of different programs in the background with this account, so I can integrate them later into the main code.

Even if everything I’m doing right now is only a tiny part of what I ultimately want to build (this is really just a microscopic piece).

Final goal of the current tests:
Make something like this work:

Example:
A post longer than 350 words (for now: 3% vote)
(word count variable depending on VP)
= eligible for curation.

Later, it could look like this:

350 words = 2% vote
500 words = 3% vote
1000 words = 5% vote
5000 words = 7% vote

Blacklisted accounts = not voted
Eligible posts = sent to a waiting list

Waiting queue:

Sorted by theme (currently works by tag, but that’s not what I want)
Sends the post to, for example:
Curation account #crypto #trading

Then applies:

350 words = 2% vote
500 words = 3% vote
1000 words = 5% vote
5000 words = 7% vote

Next idea:
Random “tips” for accounts eligible for curation.
For example: lolz, pimp, luv, pizza — within daily limits.

Some people already do this using “gif hubs” that are added to read and understand later. This looks way better than a generic automated text like the one mentioned above.

Reminder again: this account is only for testing.
No HP. No rewards. Generates nothing.

Image generated by AI

(Please don't take the questions the wrong way) Do you really know how to use your AI? Do you really know how to do prompting?

"...I want it to actually read and understand the text..."

Many do. Unfortunately, AI is just a sorting mechanism, like a sequence of scales under a conveyor belt that sorts bolts by weight. It's much more complex than that, but that's all it is. It weights text strings according to it's training data. It doesn't understand anything. It simply applies the algorithms programmed into it to the prompt it is given to string together a reply based on it's training data.

It does not understand what 2+2 means. It returns the most likely text string from it's training data '=4'.

I wanted to discuss with you the difference between machines and conscious beings. Our tools, like Hive, are never perfect. As a result of various shortcomings Hive allows automated votes. This renders machines capable of voting equal to conscious human beings, or rather degrades conscious human beings to the level of toasters. Human beings have rights. They are inherent to them, an inalienable feature of their being. When reduced to the level of toasters those human beings are essentially stripped of their humanity, that sacred consciousness no mere toaster can ever share.

An AI device has already been granted citizenship by the KSA, Sophia, a talking AI robot. While she doesn't appear particularly dangerous today, her posterity will be. Granted human rights they cannot possess, but only operate the levers of governance per their programming, AI will eventually enable their programmers to subvert democratic elections by delivering their ability to manipulate levers in voting booths per the desires of their owners, which will render sovereign people to subjugation.

I would like to hear your thoughts on this incipient existential threat we face.

Thanks!