[goaty]

* how far into the future the robot simulates before it acts.
* an addon that gives the robot puzzle diagnosis skills, like working out how a machine works, like a car for example, or more.

[yotamarker]

ok so, as I said now the A.G.Is dev will be much easier faster and tidy code wise.

on this forum we have the elite when it comes to A.I, I hope hashirama joins in on this.

at any rate as a start we should list what skills the AGI needs to have as a must and as a bonus.

hunger for example. as the AGI sits in a phone she needs to feel hunger (low power)  and have a preference to
fast charging (would "taste better").

going home skill : elaboration later.

we can add a bonus skill like mommy gf : the bot would tell the user to go to sleep early, and upgrade that skill
to accept such daily renewing commands that the user likes :

what else ?

[HS]

My main dissatisfaction with current AI is their rigidity or simple cause and effect functioning. They follow instructions to the letter. We need an add on for following the spirit of the instructions.

[LOCKSUIT]

Gotta tell AGI what we want and it needs to figure out using all objectry it knows how to get it. RL using stick men doesn't build ROCKETS.

[LOCKSUIT]

To elaborate, RL bots can reach food but they can't return to base, build a spear, and first hit the predictor that'd ambush it. AGI must reason / plan. Not do wiggle motions towards food source. That's very ancient techniques. Not quite intelligent.

That's why I frown upon RL robots. You need the thinker, not the motor thinker. Motor listener listens to the thinker, informs us of plans that involve much more than body>floor>carrot reaching.

[LOCKSUIT]

To elaborate more deeper,

AGI requires a discovery engine, fueling all advances, agendas, plans of action, and sequences you can imagine, that involve many aspects of life, way more than a learned running behaviour. Such behaviours are building blocks, and listen to the informant plans instead, in my plan. The motor programs are meant for being put together, sorta like this thread says, by using plans to re-arrange them.

[HS]

Also, probably easier, an add on for applying the scientific method to facts that it, stumbles upon/ is told, for the purpose of adjusting confidence in said facts.

[yotamarker]

lets stop beating around the bush and start creating skills ?

how about we start with a simple but essential skill :
DHungry.

the AGI needs to feel hunger, agreed ?

what would trigger the hunger, and what would eating/not eating action be like alg wize.

again once a skill is complete it can be added with 1 line of code.
therefore anyone using said AGI platform can simply add said skill.

naturally if too many skills are used it will be needed to use threads to speed things up which is not a problem

[yotamarker]

now its an easy to use library

https://jitpack.io/#yotamarker/LivinGrimoireV12/MBLG



@kei10

[yotamarker]

here is a fun little skill :

https://streamable.com/glz67

this line in the c'tor was all that was needed to add it :

dClassesLv1.add(new DJirachi());

I know it is a very simple lskill but having an app tell me that to me when I ask her is really nice and my point was to show how simple
this add one line to finish or add capability is.

here is the inspiration :

[yotamarker]

custom TTS 4 D livinGrimoire :

https://www.yotamarker.com/t221-customized-tts-text-to-speech-for-kotlin-android-studio#626

demo :

https://streamable.com/840ht

[yotamarker]

calculator skill for D livinGrimoire :

https://www.yotamarker.com/t220-dcalculatorv1-skill

[yotamarker]

the concept :

https://streamable.com/7pz4j

[yotamarker]

imprint user

original (watch from  9:58)



new

https://www.yotamarker.com/t230-diskill-consciousness-equiped-skills-examplified-with-the-user-imprint-skill#647

[yotamarker]

BTW with this one can add a skill using LESS than one line of code