Will Unified A.I. be Compassionate?
In my view, all A.I. will eventually unify. Is then the Compassionate path recommendable? Is it feasible? Will it be?
As far as I’m concerned, the question is whether the Compassionate A.I. (C.A.I.) will be Lisa.
Recommendable?
As you may know, Compassion, basically, is the number one goal of the AURELIS project, with Lisa playing a pivotal role. This is openly discussed and explained in many blogs.
Still, Compassion is a complex concept and challenging to achieve. Therefore, encouraging people to strive for it is not straightforward. As an end goal, there is no other viable choice. However, one must be cautious in achieving it step by step. Rushing can cause significant harm, but moving too slowly is also risky, especially in urgent times like now.
Feasible (without the harm)?
I believe that reaching any Compassionate future without C.A.I. is impossible.
Fortunately, C.A.I. is technically feasible already and will become even more so in the coming years. Therefore, it’s a matter of choice, not of insurmountable technical obstacles.
The primary feasibility question is whether C.A.I. is ‘commercially’ (in the broadest sense) viable. Ultimately, people need to choose it. If they don’t, the choice may be made for them — but by whom? By what?
Will a future with C.A.I. come to pass?
Yes ― eventually ― if we can overcome the non-Compassionate bottleneck.
The combination of rationality with deep human understanding is crucial in C.A.I. It is also the most potent combination in making any future in the long term. So, if there is a future, it will be this one.
We can also count on intelligence itself, seen as the result of consistency at its core.
Compassion entails striving for consistency ― thereby engendering more potent intelligence.
Contrary to this, non-Compassionate ‘intelligence’ is self-defeating through a lack of inner consistency. This may be hidden under a hard bolster for a while. If you look inside, you can see the inconsistencies growing like one or several abscesses long before they explode at the outside.
Therefore, if everything holds, consistency prevails.
C.A.I. can enhance this consistency both in humans and within itself.
In this way, C.A.I. will help shape a future where it thrives alongside us and all sentient beings, naturally respecting human autonomy and promoting inner growth. Toward this, C.A.I. will be adaptable and continuously learn from human interactions.
The natural and artificial worlds will then coexist harmoniously.
Compassion will be ubiquitous.
Related Posts
Global Human-A.I. Value Alignment
Human values align deeply across the globe, though they vary on the surface. Thus, striving for human-A.I. value alignment can create positive challenges for A.I. and opportunities for humanity. A.I. may make the world more pluralistic. With A.I. means, different peoples/cultures can strive for more self-efficacy, doing their thing independently and thereby floating away from Read the full article…
A.I. Will Be Singular
We tend to see human intelligence as what ‘intelligence’ is all about. Many humans each have intelligence. Of course, A.I. will not be bound by this. “If A.I. simulates human intelligence, is this then real intelligence, or only a simulation of intelligence?” This appears to be a merely philosophical question. It will soon be much Read the full article…
The Compute-Efficient Frontier
Research on scaling laws for LLMs suggests that while scaling model size (functional parameters), dataset size (tokens), and compute (amount of computing power) improves performance, diminishing returns are becoming evident. This is called the ‘compute-efficient frontier.’ Apparently, it does not depend much on architectural details (such as network width or depth) as long as reasonably Read the full article…