bordarretazos

“In (very) broad outline, then, the situation looks like this: It took a few million years for biological evolution to result in Homo sapiens, the first species to engage extensively in abstract thought and symbolic reasoning. Humans are the cognitive species par excellence, and we have used these abilities to instantiate symbolic reasoning and logic in artificial media, which now have their own evolutionary dynamics. In suggesting a parallel between biological and artificial evolution, I do not intend to minimize the significant differences between them. These differences can be understood through what I call the two great inversions. The first inversion replaced the biological mandate “survive and reproduce,” the process through which biological evolution proceeds, with the computational mandate “design and purpose.” Computational systems, unlike biological organisms, are designed for specific purposes, so their evolution in this sense proceeds in top-down fashion.

The second great inversion concerns the directionality of evolutionary trajectories. From day one, biological organisms had to be able to cope with fluctuating and unpredictable environments; if they could not, they did not survive. From this foundation, a cognitive trajectory toward increasing complexity eventually led to the capability for abstract thought. Computational media, by contrast, instantiate abstract thought at a foundational level in the logic gates, which are almost completely deterministic. From this base, layers of software increasingly enable computational media to deal with uncertain and ambiguous information. When sensors and actuators are added, computational systems may also be able to deal with environmental ambiguities and fluctuations. In addition, developments beyond von Neumann architectures, such as neural nets and neuromorphic chips, further advance in the direction of fault-tolerant systems able to draw inferences from unruly data and to operate successfully even with contradictory information. In this sense, then, the trajectory of computational evolution proceeds in the opposite direction from biological evolution. The biological progression is from uncertainties up to abstraction, which the computational progression inverts so that it proceeds from abstraction up to uncertainties.

In summary, my focus on cognition is intended to capture the aspects of computational media that make them not just another technology but a cognitive technology able to interpret, disseminate, and contextualize flows of information so they become meaningful within specific contexts. This is why these media are able to impart to a huge variety of other technologies the advantages that cognition bestows -flexibility, adaptability, and evolvability. Once mechanical processes are interpenetrated by computational components able to interpret information and create meaning from it, they are able to carry out tasks impossible for machines without cognitive abilities.

For example, mobile robots stacking boxes in a warehouse are able to do tasks that would be impossible for a forklift lacking a human driver. The forklift by itself cannot navigate around obstacles; it is unable to judge properly how to lift and stack boxes, nor does it know when to start and stop. The forklift can operate only when human cognition comes to the rescue and provides the cognitive abilities that it lacks. But as the box-stacking robots demonstrate, incorporating computational components into machines now enables machines to take on tasks that formerly could be done only by humans.”

Katherine Hayles, Postprint, pág. 11