The Eve of Intelligent Design

The Eve of Intelligent Design

The foundation of most current AI is neural networks, which approximate how humans and all other animals "think".  While this approach has resulted in some amazing technologies that begin to approach true reasoning, there are two large issues: we are emulating a biological system using digital computers, and that biological system is the result of billions of years of happen-chance evolution and is far less than optimal.  As Darwin said, the appearance of design is illusory.

Emulating a sub-optimal biologic process on digital computers is likely resulting in highly inefficient AI systems.

I believe modeling intelligence using neural networks is a stop-gap measure until we have true Native Artificial Intelligence.  Native AI will occur when the perceptrons of neural networks are replaced with higher order functional programming nodes.  These nodes will likely be wired together using familiar model training and fine-tuning techniques.

The first generation of Native AI will be linear, with fixed inputs, then inference, and an output. Recursive Native AI, with continuous input, inference, feedback, and output will result in parity with what a digital computer could accomplish with optimal native code.

We are the Intelligent Designers

The emergence of digital intelligence would not have happened without us.  We are witnessing a jump in evolution, as it transitions from point mutations in DNA tested over millennia, to intelligently designed architectures trained using intentional data in mere days.

The next 20 years is going to be a roller-coaster of achievements and challenges, and the most amazing period ever to be alive.