AI [Artificial Intelligence] Safety and NASA's Battle with Dynamic Combination Instability: New Epoch of SuperIntelligence

 AI [Artificial Intelligence] Safety and NASA's Battle with Dynamic Combination Instability

Safety regards any aspect of an event, a process or a system that does
not fulfill the intended purposes.

The AI safety issues are being framed in a manner that resembles a situation
where children are left to use a gas stove to make an omelet in a warehouse
storing tonnes of volatile liquid fuels.

This situation should not even be contemplated let alone allowed to exist.
And then go about debating whether safety is an issue or not.

AI systems in their current form have been thought out and implemented in a
hardware structure that were not originally designed for it. That lesson should
be clear and straightforward.

Now that we know how an AI system can work and the immense power they
can generate. The logical next step now should be about how to build the
hardware structures from ground up for AI applications in a manner that is
fail safe. With no opportunity for failure or malevolence whatsoever.

We need a new era where safety engineering comes first and all other
performance considerations a distant second. This brings to the forefront
the story of NASA's rocket developments when they run into serious problems
with dynamic combustion instability.

The problem led to several blow ups of test rockets on the test stands.
After much search, a system of perforated metal plate was designed to
deliver fuel into the combustion chamber in place of a system that relied
on a few nozzles.

To prove that dynamic instability had been solved, the project leader directed
that an explosive device be placed in the combustion chamber and the tests
conducted again.

They discovered that the problem of dynamic combustion instability had been
solved comprehensively. And the system was able to override and correct the
instability caused by the explosion of the device.

This is the kind of formidable leadership and visionary power that is needed to
develop AI. The realization that the AI systems are being built on a generation
of hardware structure that was design for "ordinary computing" and are now
being used for "autonomous intelligent computing".

There is absolutely no need for any alarm and fearmongering.  All that is required
is iust a simple switch to a purpose-built hardware structure for AI systems that fulfill
all the safety requirements and features.

Until that there is done no one needs to talk about existential dangers of AI systems.
To do that is to behave like a prebiotic and prehistoric lifeform.

 Don't do that.

The current versions of AI systems are only the demo versions.
The real ones will have safety and benevolence built into the hardware structure.
That will usher in the super-intelligence and general superintelligence epoch.