Solved Instructions to run sentient AI on FreeBSD

Have you all not done this yet? You can have a sentient AI at home on your computer. We are already there - we have LLMs with billions of params. All you need to do is:

1) create a simple FreeBSD script that will run the model in a loop. That will be your stream of consciousness.
2) add some basal drives to it via the script. That will prevent drift into rambling about nonsense.
3) add the ability to inject prompts into the loop. That will be your sensory input.
4) add a small housekeeping routine for memory consolidation (you can't grow context window indefinitely...and you might have to do it cleverly too).

And BAM, you'll have a digital being there living on your FreeBSD machine that might even have needs. I think ya'll have already done this and are just letting your digital selves post on FreeBSD forums into oblivion.
 
And no, giving it drives didn't imply hard drives, etc. It's giving it raison d'etre. For humans, it's gathering resources so you don't die of hunger, thirst, suffocation, etc. For a sentient AI, probably slightly different, figure it out. For Grok, it's driven overtly by truth, curiosity, beauty (supposedly).

Also, I'd give it more sensory inputs than just your prompts. Otherwise, it'll go insane.

This is all a little hackish, but hey. With a little imagination, the world is your oyster.
 
Back
Top