‘Why I believe fantasy is harming robotics'’

Ex Machina: Alicia Vikander — artificial intelligence in the award winning film.
by Jaron Lanier

Considered one of the most brilliant minds of his generation, his interests include computer science, philosophy, art and classical music. He is considered a founding father of virtual reality.

A computer has beaten a grandmaster of the hyper-complex board game Go. Despite this, argues Jaron Lanier, technology companies are prone to fantasies about artificial intelligence.

Artificial Intelligence (AI) is, more than anything else, a funding category for research. It incorporates a wide range of disciplines and pursuits, that were bundled together by historical accident in many cases.

AI steps on its own foot periodically. There is a crazy, wild-eyed: ‘We are about to understand how to replicate a person’. Then the funders say: ‘You did not deliver’. There is an AI winter and you watch your graduate students having their careers ruined. Then it happens again.

I am not anti-AI algorithms. I make them, and am fascinated by them. But there has to be a division between the engineering and science, and the storytelling about it, the fantasy life of it — perhaps the religion of it. If we could just be good engineers and scientists, we would free ourselves from this burden of constant self-destruction.

“In order to have the fantasy that this thing is a free-standing creature, we are pretending people don’t exist”

It would be crazy to argue there is not promise in pursuing science information systems and sensing systems, and algorithms to understand it. We just observed gravity waves for the first time. Does that mean we will suddenly have anti-gravity devices? Maybe some day.

But we have a field that has lost its moorings to fantasy. As an engineer, if I say: ‘I am making this algorithm into something intelligent, or conscious’, I can’t define those terms.

I frequently review student work, and some say: ‘I have made an intelligent system’. For instance I saw a machine that makes beverages. It had fancy software, but people found it hard to use. Nobody cares how fancy your algorithm is. The only thing you can measure is how well the machine works, in the end.

Engineering without concrete grounding in reality becomes dangerous. If a drone goes around killing the wrong people, who cares whether it did so because of an intelligent algorithm that made the wrong decisions, or just because it was malfunctioning? The difference between one machine that works terribly and another machine that works terribly is not important.

There is an economic angle. I love that you can go online and get something converted to German automatically. My own lab does that. We have real-time Skype translation now. But we do it by scraping together the efforts of millions of translators who do not even know what is happening to them to get the examples.

In order to have the fantasy that this thing is a free-standing creature, we are pretending these people don’t exist. We are potentially creating a massive wave of unemployment that does not need to happen. Acknowledge that people are just contributing in new ways. Their Go games are informing a Go algorithm. Pay those people.

This fantasy of these artificial creatures makes us ignore our own contributions. If we fall into this mythology from all these screenplays and productions, if we accept the machines as being living things, unfortunately we will say: ‘It is the machine’s fault. The machine did this. The machine was good or evil.’

At the moment people accept the notion they have less responsibility, we start losing a bit of civilisation. We lose a bit of society. We lose a bit of ourselves, and there is no reason for it.

Believe in yourselves as real, mysterious, living organisms who are not yet fully understood by science. We live in a sea of mystery. We understand so little of our situation. Take joy in that mystery. In that mystery you also find the profound sense of responsibility that you are morally impelled to hold onto.

Science fiction is great in the theatre. It stinks out on the street.

You Decide

  1. Should technology companies invest heavily in researching artificial intelligence?


  1. Write a list of five questions you have about artificial intelligence. Then give your questions to a partner, who needs to do some research and then give you a 30-second explanation of the answer to each one.

Word Watch

AI winter
A period when there is less interest in AI and therefore funding for research in it is cut.
A process or set of rules which are followed in making a calculation.
Gravity waves
Last month, physicists from the LIGO institute in the USA revealed that gravity moves in waves. The discovery could help to reveal some of the secrets behind the creation of the universe.
One computer scientist said last month that AI could leave half of the world’s population unemployed within 30 years.
A 3,000-year-old game from China which involves two players alternately placing pieces, called stones, on a board. Last week AlphaGo, a program created by Google’s artificial intelligence company DeepMind, completed a 4-1 win over Lee Sedol, a South Korean professional who is ranked second in the world at Go.
Artificial intelligence has featured in several Hollywood blockbusters. Recent films on the subject have included Ex Machina and Transcendence. But its use is not new; in 2001: A Space Odyssey (which was created in 1968), HAL 9000 was a sentient computer.