Robot Evolution? How the Trick Is Done
It’s been decades since Richard Dawkins committed the Weasel Blunder and since Tim Berra committed Berra’s Blunder, but some evolutionists still don’t get it. You can’t design something for a purpose and call it Darwinism. Even if some randomness is thrown in, once a goal is specified in advance, that’s not evolution; it’s intelligent design.
An example comes from PLOS ONE: “Morphological Evolution of Physical Robots through Model-Free Phenotype Development,” by Brodbeck, Hauser, and Iida. Look for the evidence of guidance by the investigators:
Artificial evolution of physical systems is a stochastic optimization method in which physical machines are iteratively adapted to a target function. The key for a meaningful design optimization is the capability to build variations of physical machines through the course of the evolutionary process. The optimization in turn no longer relies on complex physics models that are prone to the reality gap, a mismatch between simulated and real-world behavior. We report model-free development and evaluation of phenotypes in the artificial evolution of physical systems, in which a mother robot autonomously designs and assembles locomotion agents. The locomotion agents are automatically placed in the testing environment and their locomotion behavior is analyzed in the real world.This feedback is used for the design of the next iteration. Through experiments with a total of 500 autonomously built locomotion agents, this article shows diversification of morphology and behavior of physical robots for the improvement of functionality with limited resources.
Stamped by Design
These researchers from Switzerland carefully crafted a “mother robot” that could assemble pre-fab parts into blocks that could perform some simplified locomotion. The algorithm was set to reward “offspring” that performed faster. Despite employing Darwinian words like “fitness” and “selection,” their work has “design” stamped all over it. Reporters, though, went ape attributing this to Darwinian evolution.
At Phys.Org, Sarah Collins titled her report, “On the origin of (robot) species” in tribute to the Great Bearded Guru. She’s like a viewer of a magic show unaware of how the trick is done:
For each robot child, there is a unique ‘genome’ made up of a combination of between one and five different genes, which contains all of the information about the child’s shape, construction and motor commands. As in nature, evolution in robots takes place through ‘mutation’, where components of one gene are modified or single genes are added or deleted, and ‘crossover’, where a new genome is formed by merging genes from two individuals.
In order for the mother to determine which children were the fittest, each child was tested on how far it travelled from its starting position in a given amount of time. The most successful individuals in each generation remained unchanged in the next generation in order to preserve their abilities, while mutation and crossover were introduced in the less successful children.
In all fairness, Collins points out some differences between robots and natural organisms. Still, nowhere does she explain how the experiment clearly requires design instead of natural selection. She thinks the “mother” robot did the designing. And she takes on faith the opinion of one of the authors that they were watching Darwinian evolution happen before their eyes:
“Natural selection is basically reproduction, assessment, reproduction, assessment and so on,” said lead researcher Dr Fumiya Iida of Cambridge’s Department of Engineering, who worked in collaboration with researchers at ETH Zurich. “That’s essentially what this robot is doing — we can actually watch the improvement and diversification of the species.“
“Learning to Evolve”
At the BBC News, evolution reporter Pallab Ghosh titled his coverage, “Robots learn to evolve and improve.” It’s not clear how one could “learn to evolve” if evolution is an unguided natural process, but that’s not the only conundrum in his article. Like Collins, he fails to make any distinction between intelligently designed robots and natural processes.
Engineers have developed a robotic system that can evolve and improve its performance.
A robot arm builds “babies” that get progressively better at moving without any human intervention.
The ultimate aim of the research project is to develop robots that adapt to their surroundings.
There’s a second conundrum: if human minds are developing robots that evolve, isn’t Ghosh making a case for intelligent design?
What the designers and reporters all seem to be missing is the fact that goals were determined from the outset. “Improvement” was defined as the ability to move faster. Yet in nature, not every successful animal is the speediest (consider the sloth, or the fabled tortoise and hare). Darwinian evolution cannot work toward a distant target. As Paul Nelson remarks in the film Living Waters, “Any evolutionary process you consider, any materialistic process you can consider has no foresight. It can’t see five years, five seconds, five milliseconds into the future. For that, you need a mind.”
An Aim and an Approach
The paper actually makes a powerful if unintended case for intelligent design when you think about it. Ghosh reports that Dr. Iida got into robotics because the ones he saw in real life were not as good as the ones he enjoyed in movies like Star Trek and Star Wars. “His aim was to change that,” Ghosh says, “and his approach was to draw lessons from the natural world to improve the efficiency and flexibility of traditional robotic systems.” He had an aim. He had an approach. He wanted to gain knowledge, or information. So he looked at the efficiency and flexibility of natural solutions, where he found efficient designs worth copying. In other words, he was motivated by biomimicry — an approach saturated with design thinking.
But he used an evolutionary algorithm, someone might complain. True, but it wasn’t evolutionary in the Darwinian sense. There’s no such thing as a Darwinian “algorithm” despite the use of familiar lingo like mutation, selection, and fitness in the paper. Algorithms are intelligently designed for function. Once an algorithm is defined, a mindless mechanism like a computer program or robot can use it, applying inputs and monitoring outputs, as in this case. But those mechanisms were also predesigned to implement the predetermined goal.
Animal and plant breeders use “evolutionary algorithms” of a sort; they know what they want; they use algorithms of sexual reproduction, and they monitor the output to decide what offspring get to breed in the next iteration. All this is under the guiding hand of the intelligent agent (the breeder). Artificial selection is intelligent design, not Darwinism.
If the authors and reporters really wanted to see materialistic Darwinian processes in action, they should have taken their hands off the equipment, shut the door, and let nature take its course. Most likely, nothing more interesting would happen than rust.
Two Blunders in One
Dawkins set a goal of generating Shakespeare’s phrase “Methinks it is like a weasel.” Berra watched cars “evolve” but missed the role of designers. The designers of these robots (and their Darwin-friendly reporters) committed both blunders. They had a target, designed a way to reach it, yet presumed after the fact that their carefully engineered “mother robot” resembled a mindless, material entity working by Darwinian natural selection. We can at least thank them for providing another opportunity to show why they really made the case for design.