Why Do We Fear Robots?

By Kai Goerlich and Christopher Koch

Normally, no one fears computers. Indeed, many of us love them and can’t imagine our lives without them.

But if that box sprouts mechanical arms and legs and a head, our attitude changes. The closer that robots get to a human or animal form, the more we begin to feel the mixture of fear and empathy that has defined our relationship with our fellow beings since we first began dragging our knuckles around millions of years ago.

If we skew the design of these machines toward empathy, we get Pepper, which can identify the principle human emotions – joy, sadness, anger, and surprise – and adapt its behavior to accentuate the positive and comfort the negative.

If we let fear guide the construction process, we get autonomous tanks, drones, and early prototypes of human-like warriors – not to mention fictional machine menaces like The Terminator that have haunted our imaginations since the beginning of the Industrial Revolution.

It’s fair to say that over the course of history, we imagine robots being more like the Terminator than Pepper. That’s because we have a tendency to anthropomorphize anything that vaguely resembles us. We’ve already created machines that are faster and stronger than human beings. The possibility that they may also one day become more capable than us can be terrifying.

Now that Artificial Intelligence (AI) has matured to the point where machines can simulate human capabilities such as language, vision, and hearing with a high degree of accuracy and even learn on their own, it seems that our worst fears may soon be realized. To be sure, the threat of robots taking significant numbers of jobs – especially if those jobs are rote and repetitive – is real.

But just because robots and AI increasingly look and act like humans, that doesn’t mean that they will gain superiority over us. We must distinguish between weak AI, which is about mimicking specific human abilities, and strong AI, which is the idea that machines will one day cross over the line to human-like consciousness.

It will be quite some time – if ever – before machines and robots become capable of thinking in the ways that humans can. Robots and machines are completely reliant upon data and algorithms supplied by humans. They can’t improvise or be creative like we can.

This video compilation of epic robotic fails should reassure us that we are still far away from truly intelligent robots. As Reinhard Karger from the DFKI (German Center for Artificial Intelligence), puts it, robots are really bad at what humans are good at and vice versa.

A short list of skills in which we can knock the rivets out of our machine-based brethren: childcare, coaching, counseling, cooking, writing (that requires creativity or new insights), art, medicine (as long as human doctors work on their empathy skills), and tour guides (just don’t tell your customers that a sunset is merely a collection of colors as a robot might).

On the other hand, let us crunch numbers manually or do repetitive tasks and you’ll quickly see how overwhelmed or bored we become and how good software and machines are already. Even without much computing intelligence, robots have been replacing humans on assembly lines for decades now.

And in more complex situations that call for excellent memorization or pattern-matching skills, such as chess or GO or Jeopardy, humans are no longer any match for machines.

But let’s not be distracted from the fact that these achievements, while remarkable, are still merely examples of task mastery rather than human-style, holistic thinking. As Andra Keay, Managing Director of Silicon Valley Robotics, put it during the recent Singularity University conference in Berlin, robots are good at tasks and bad at jobs.

Many of the jobs that are threatened by automation are actually tasks, or groups of tasks. And here, Karger and Keay share a view: many of these task-oriented jobs are boring or risky and humans would be better off giving them up to robots.

Karger and Keay believe that robots should take over the “Four-D” jobs:

• Dirty: cleaning, sewage, garbage
• Dull: proofreading, invoice verification
• Dangerous: construction, mining, heavy machinery
• Deadly: cleaning areas with poisonous materials

We are at the dawn of smart things becoming part of our lives. Like any technology, machine learning and robotics can be used for good or bad. Let’s make sure that we use these tools to free up our capacity to be human and to improve our lives.

This blog is the sixth in a six-part series on machine learning. It first appeared on Digitalist Magazine