How to Build a Neuron: Exploring AI in JavaScript Pt 1
Eric Elliott

Ughh. Here we go again. The brain as computer or computer program argument. Despite having been fundamentally dismantled as a viable analogy for understanding how the brain works by a huge array of notable philosophers, neuroscientists, and cognitive scientists over the past 15–20 years it still won’t seem to die. Mostly it seems that technologists or programmers or engineers still cling to this dying and disproven concept. In silico neural networks are not representative of neural processes in the human brain. When viewed in the most favorable light, they attempt to mimic such processes, and do so poorly. When viewed realistically they completely fail to achieve even a basic output of the functions the human brain is capable of. I am not naive and recognize that advances in computer science will most likely continue to narrow the gap between the abilities of the human brain and machines but at present that gap is vast. I would suggest that gap can never be completely crossed without some future, currently non existing technology, that allows fusing of biological processes with technology.

In the present day computer neural networks are capable of generating “cool and trippy” images like those of Googles deep thought project. They have also proven useful in certain data analysis applications. Maybe someday they might form the basis of a rudimentary AI. Despite claims to the contrary current AI systems are totally lacking in even the most basic forms of human intelligence. They are extremely sophisticated computers and can compute with uncanny speed and accuracy. For many people this is enough to suggest the beginnings of intelligence at the very least. For those who have studied these issues for a very long time they are only an illusion.

Show your support

Clapping shows how much you appreciated Daniel DeMarco’s story.