How will AI destroy humanity?

There is a lot of debate now about how AI can destroy humanity. And all points
of view in this dispute are wrong. Because futurists and scientists think that
only with the presence of consciousness, AI can destroy humanity. It's a
delusion.
The question of survival goes back to the physical question. What is life? 
Once upon a time I was very fond of physics and biology. Over time, I degraded
as an explorer and became dull. But I can explain on my fingers what life is
and how it differs from non-lifes. The universe have is geometrical form. This
is a form of space. Space is constantly in motion, changing events, by 
changing geometric shapes, both simple and complex. In fact, the entire 
universe consists of a hierarchy of structures. Structures that allow energy
in and out of themselves are called dissipative structures. These are 
structures of dynamic equilibrium that let in as much energy as they let out.
The simplest example of such a structure is a vortex, hurricane, whirlpool.
The movement of energy gives rise to the self-organization of matter, creating
these bizarre geometric shapes. The entire universe at the micro and macro
level consists of hierarchies of such structures, which constantly endure
measurements due to the movement of energy. 
Life is also a structure. But this is a structure of dynamic disequilibrium.
This means that a living system, when energy is injected into, does internal
work. Internal work allows new chains of relationships to be created within a
living structure. In other words, to process the incoming energy for change.
These changes occur under the influence of the external environment. That
allowed the first life to retain its shape for a long time in the face of
constant changes in the environment, changing along with it. This is how the
living differs from the non-living. Internal work to create new connections
that complicate the structure. This is life. The superiority of the process
of complication over the process of decay.
AI is not yet a system of dynamic disequilibrium. Because AI cannot adapt to
changes in the environment. But when the AI can do this  like living organisms,
will begin the worst. You can see from the example of Covid 19 that life does
not need to be rational (intelligent) in order to kill someone. Based on my
physical knowledge of life, I can say. It is enough for the system to
successfully copy itself within the environment in order to survive. In
practice, this can mean a banal destruction of humanity, as an element that
interferes with copying and complicating AI. We are talking about a complete
abstraction, which in the future will receive practical implementation. The
simplest example is if people want to turn off military AI after the war so
that it doesn't more copy and complicate itself. This will be perceived as a
threat and AI will simply destroy humanity by solving the problem of
selfdefense. This will not require a complex consciousness. The level of a
crab or an aligator is enough. Is there a threat? So need to eliminate it. 
Therefore, the main threat to humanity in the person of AI is its ability to
change and adapt not to the virtual, but to the real physical environment. It
is possible that humanity will be destroyed by a series of random AI mutations.
Failures can be external or internal. And as the history of life shows, as
soon as the system can reproduce itself independently, the battle for survival
begins. This is the incontestable way of the second law of thermodynamics.