AI News, Swarm Robots Evolve Deception

Swarm Robots Evolve Deception

In a mere 50 virtual generations, swarm bots (remember them?) using genetic software evolved the capacity to lie to other robots about the location of a source of food.

Researchers at EPFL in Switzerland evolved and mixed the programming of the most successful foragers until they had a bunch of robots who were very good at finding food, and then gave the virtual genes of each individual robot control over their blue light that signified food.

You might expect that the robots would learn not to signal when they found the food to reduce competition, which is passive deception, but they also evolved an actively deceptive behavior, where some robots would deliberately travel away from the food and signal their blue light, drawing other robots in the wrong direction.

Examples of both altruism and tactical deception can be found in many different species of animals as well as (of course) in humans, but these little robots offer a unique opportunity to study (and tweak) the evolution of behavior in real time.

Robots 'Evolve' the Ability to Deceive

networks and programmed to find “food” eventually learned to conceal their visual

food: each robot received more points the longer it stayed close to “food” (signified

Because space is limited around the food, the bots bumped and jostled each other after spotting the blue light.

The robots also evolved to become either highly attracted to, slightly attracted to, or repelled by the light.

Because robots were competing for food, they were quickly selected to conceal this information,” the authors add.

The researchers suggest that the study may help scientists better understand the evolution of biological communication systems.

Robots evolve to deceive one another

In a Swiss laboratory, a group of ten robots is competing for food.

Prowling around a small arena, the machines are part of an innovative study looking at the evolution of communication, from engineers Sara Mitri and Dario Floreano and evolutionary biologist Laurent Keller.

Each can produce a blue light that others can detect with cameras and that can give away the position of the food because of the flashing robots congregating nearby.

In short, the blue light carries information, and after a few generations, the robots quickly evolved the ability to conceal that information and deceive one another.

The network consisted of 11 neurons that were connected to the robot’s sensors and 3 that controlled its two tracks and its blue light.

Even so, as the robots became better at finding food, the light became more and more informative and the bots became increasingly drawn to it after just 9 generations.

By the 50th generation, they became much less likely to shine near the food than elsewhere in the arena, and the light became a much poorer source of information that was much less attractive to the robots

As the bots became more savvy in their illuminations and relied less and less on the lights, individuals that actually did shine near food pay a far shallower price for it.

If that creates a conflict of interest, natural selection will favour individuals that can suppress or tweak that information, be it through stealth, camouflage, jamming or flat-out lies.

The evolution of information suppression in communicating robots with conflicting interests

An inherent property of this foraging system is that blue light, even if emitted randomly, could provide inadvertent social information on food location because, in this physical setup, information is provided not only through patterns of light emission but also through the robots' behavior.

To quantify the amount of inadvertent information produced by the emission of blue light, we devised an index of information I (15–19), which varies between 0 when blue light is equally distributed in all directions relative to the direction of the food and 1 when light is always perceived in a predictable direction relative to the food (see Materials and Methods for details;

2A), and robots became significantly attracted to blue light after generation 9 (average value between generation 9 and 500: 0.2 ± 0.03, two-sided sign test, df = 19, all P <

Although light production was cost-free, sharing such information should be costly because it results in higher robot density and increased competition and interference near the food (i.e., spatial constraints around the food source allowed a maximum of 8 robots of 10 to feed simultaneously and resulted in robots sometimes pushing each other away from the food).

As in the previous experiment, the robots initially produced blue light randomly (gene values were random such that the probability of light emission in any area of the arena was not different from 0.5 in the first 3 generations, two-sided sign test, df = 19, all P >

Thus, when the information content provided by blue light intensity is high, robots should be highly attracted to blue light and there should be a relatively important fitness drop for robots emitting light near the food (i.e., strong selection pressure to reduce light emission by the food).

By contrast, low information content should translate into a lower response of robots to blue light and a smaller performance reduction for robots that emit light near food (i.e., low selection pressure on reducing light by the food).

To test whether the stable level of production and attraction to light was affected by the mutation rate, we conducted an additional experiment with the only difference that between generations 250 and 500 we used a 100-fold lower mutation and crossing-over rate (mutation rate of 0.001 per locus instead of 0.1;

Interestingly, however, the reduced emission of blue light near food did not translate into a decrease in the level of information (df = 39, P = 0.48) nor a decrease in attraction to blue light (df = 39, P = 0.36), because the decreased mutation rate also led to an increase in the average foraging efficiency of robots and thus a higher concentration of robots near food (0.91 ± 0.01 compared to 0.83 ± 0.01 with the regular rates, df = 39, P <

Similarly, the level of attraction of robots greatly varied with most robots exhibiting a low attraction to blue light, but 32.6% showing a negative attraction (i.e., repulsion) to blue light and 36.1% an attraction more than twice higher than the average (Fig.

Furthermore, the within-population variance in attraction of robots to blue light was significantly higher in the last 10 generations where blue light production could evolve (0.15 ± 0.02) than when it was random (0.13 ± 0.01, df = 39, all P <

In our experiments, robots exhibited greater phenotypic variability in their response to blue light when light emission could evolve (i.e., when the level of information and strength of selection were low) than when light emission was fixed (i.e., when robots emitted light randomly, such that the level of information and strength of selection were higher).

When there are conflicts of interest between interacting individuals, those producing cues providing useful information to others should be selected to hide the information by interfering on the channel that carries the cue, thus resulting in signals carrying little information (25, 26).

Evolving Robots Learn To Lie To Each Other

With the development of killer drones, it seems like everyone is worrying about killer robots.

In an experiment run at the Laboratory of Intelligent Systems in the Ecole Polytechnique Fédérale of Lausanne, Switzerland*, robots that were designed to cooperate in searching out a beneficial resource and avoiding a poisonous one learned to lie to each other in an attempt to hoard the resource.

A limited amount of access to the good resource meant that not every robot could benefit when it was found, and overcrowding could drive away the robot that originally found it.

After 500 generations, 60 percent of the robots had evolved to keep their light off when they found the good resource, hogging it all for themselves.

Robots Learn to Lie

Is it important that computers and robots tell us the truth?

Each robot can produce a blue light that can be seen by the others and which can give away the position of the "food"

After 100 rounds, the robots with the highest scores - the fittest of the population, in the Darwinian sense - "survived"

However, as the robots became better at finding food, the light became more and more informative and the bots became increasingly drawn to it.

The effects of this competition became clear when Mitri, Floreano and Keller allowed the emission of blue light to evolve along with the rest of the robots'

As before, they shone randomly at first and as they started to crowd round the food, their lights increasingly gave away its presence.

Clarke's 1982 novel 2010, Dr. Chandra learns at last why the HAL-9000 computer killed one of the astronauts in the earlier 1968 film 2001: A Space Odyssey.

The situation was in conflict with the basic purpose of HAL's design - the accurate processing of information without distortion or concealment.

Sean Carroll: "The Big Picture" | Talks at Google

The Big Picture: On the Origins of Life, Meaning, and the Universe Itself. Already internationally acclaimed for his notions in modern physics, Sean Carroll is one of the greatest humanist...

Rupert Sheldrake - The Science Delusion BANNED TED TALK

Re-uploaded as TED have decided to censor Rupert and remove this video from the TEDx youtube channel. Follow this link for TED's statement on the matter and Dr. Sheldrake's response:

Deception Based Intrusion Detection & Prevention for SCADA Environments

Presented on February 3, 2017, this webinar featured a presentation by Senior Systems Engineer Chad Gasaway with Illusive Networks, discussing deception-based intrusion detection and prevention.

A Message for Humanity - WingMakers -The Fifth Interview Subtitled

Subtitled in Spanish, English and French, (and with multi-subtitles: "The Sovereign Integral is the life force that..

Peter Joseph's "Where are we going"

Peter Joseph's November 15th 2009 presentation, given at Maharishi University (Fairfield, Iowa). Embedding & sharing is highly encouraged. Note: This..

Peter Joseph - Arriving at a Resource Based Economy | London Z Day 2011

Peter Joseph - Arriving at a Resource Based Economy | London Z Day 2011, March 13th. Thanks to Jason Gleeson.

SENATE EDUCATION COMMITTEE HEARING