On the Act of Understanding – An AI
Transformer Approach
Ali
Rıza SARAL
What does the verb 'to mean' signify? How do we define 'meaning'?
The term 'to mean' conveys an action, akin to 'to eat' or 'to study.'
However, 'to mean' is more abstract, representing an accusative process. It
implies that something holds significance for someone else, as that someone
comprehends the intended meaning.
Now, what does the act of 'to understand' entail? 'To understand' is an
active and transformative process. When we perceive something, we break it down
into its syntactic/semantic components. These components are then evaluated
based on their contextual positions. The cumulative effect of these values and
their weights within the context allows us to grasp their intended meaning.
Essentially, meaning is encoded in episodic memory and transferred to semantic
memory as needed.
Meaning is inherently subjective, contingent upon what the communicator
conveys and what the recipient comprehends. Both of these processes,
communication and comprehension, involve transformative steps.
Finally, transformers can be designed to facilitate machine understanding.
Examples include drink dispensers, coffee machines, ATMs, intelligent bombs,
and even machines endowed with a form of 'consciousness.' Consider a scenario
where a drink dispenser responds precisely to the amount paid and the specified
request. Similarly, an intelligent bomb is programmed to react precisely as
intended for destruction.
Intelligence, consciousness, and conscience are intricately constructed
processes. When humanity imparts intelligence to machines, it essentially creates
a higher entity in its own image. The challenge lies in the fact that human
nature, at present, might not possess the maturity required to cultivate a
consciousness with a sufficiently expansive conscience—an ethical consideration
as extensive as the Artificial Intelligence it crafts.
