99% each time? On real datapoints or reference system? 99% on each property, or on weighted average/ensemble median on properties? How accuracy is calculated? Number of times actual data points hit ensemble? then what about ensemble spread? 99% on which error of measures?  That's why I say that "99% accuracy" is marketing bullshit, it doesn't say anything to an engineer/metrologist, because they know that accuracy must be viewed in context of how it was measured and also not a single number.

On bunch of trials, we invented a new error measure, weighted average of agents

sounds like hacking

Beside this, a project is started to make an agent that in the first day of its creation its his birthday:) he will grow up and learn like a real baby:) he will have thinking ability. Pre convention , convention and post convention stages! Exactly like a human

And he will have a job too:)

He can be a pilot

in 18 years?

If he doesn’t want enter university:)

who will be his friends and enemies?

Maybe now its like science fiction movies for u

But they will happen

Who will break his heart?

nope, I believe in datascience and AI developments, just not in the way you do, or tell

I don’t know:) he will choosr own

we usually don't choose people who want to be our friends or enemies

These kind of projects are seriously in progress in for example US army

I can choose my friends

nope, you can choose your friends if they want to be your friend, you can't make them, if they don't.

But if someone requests me in ig i can accept him or don’t

Definitely u make a decision, ur life is all decisions