The Deep Learning A.I. Playbook – Strategy for Disruptive Artificial Intelligence by Carlos E. Perez

I am reading The Deep Learning A.I. Playbook – Strategy for Disruptive Artificial Intelligence by Carlos E. Perez. The author is (much like the people here on the forum) frustrated by how loosely people use the terms AI and deep learning. He presents an interesting perspective on AI as an evolutionary tree, which begins in 1955, the year the term artificial intelligence was coined.

The first approach that began to dominate was symbolic AI, or GOFAI (Good Old Fashioned AI). This includes expert systems and deductive logical systems, that is, systems based on symbolic logic. This direction has one fundamental drawback: it cannot effectively learn from data. It requires developers to sweat out symbolic rules. In Carlos's analogy, symbolic AI is likened to the emergence of plants.

Next comes the era of reptiles. Reptiles are compared to the development of statistical methods, such as Bayesian graphical models and machine learning techniques. This era introduces a new approach, in which the system is already capable of adapting to data.

Finally, in 2012, we enter the epoch of deep learning, which significantly builds on ideas from the late fifties. If we continue with Carlos's analogy, we are talking about mammals, a sort of evolution from reptiles with distinctly new and expanded capabilities.

How do you like this analogy?