Analyzing the Contextual Shortcomings of Artificial General Intelligence. (arXiv:2304.00002v1 [cs.AI])

Even in the most cutting-edge Artificial General Intelligence (AGI)
endeavors, the disparity between humans and artificial systems is extremely
apparent. Although this difference fundamentally divides the capabilities of
each, human-level intelligence (HLI) has remained the aim of AGI for decades.
This paper opposes the binarity of the Turing Test, the foundation of this
intention and original establishment of a potentially intelligent machine. It
discusses how AI experts misinterpreted the Imitation Game as a means to
anthropomorphize computer systems and asserts that HLI is a red herring that
distracts current research from relevant problems. Despite the extensive
research on the potential design of an AGI application, there has been little
consideration of how such a system will access and ingest data at a human-like
level. Although current machines may emulate specific human attributes, AGI is
developed under the pretense that this can be easily scaled up to a general
intelligence level. This paper establishes contextual and rational attributes
that perpetuate the variation between human and AI data collection abilities
and explores the characteristics that current AGI lacks. After asserting that
AGI should not be seeking HLI, its current state is analyzed, the Turing Test
is reevaluated, and the future of AGI development is discussed within this
framework.