Rescuing the Future | AI | Isaac Asimov’s Robots and Empire | a book from the past addressing our future

Given the worried attention Artificial Intelligence is getting so much of now, I decided to re-read my favorite of all Isaac Asimov’s robot stories, his 1985 novel Robots and Empire, to see if it can help us.

You don’t read Asimov for the human characters, you read him for his daring, intelligent and often beautiful ideas.

The human characters can be tiresome. The famous Harry Seldon of the Foundation trilogy, who develops the science of psychohistory that allows him to manage, with difficulty, humanity’s galactic empire, is the ultimate know-it-all. He can get on your nerves.

But there is one set of characters I have always loved – the robots.

But do robots in a novel published 40 yrs ago have any relevance to what is happening now? Well,  Asimov may have been the first person to ever address AI safety. With his famous ‘Three Laws of Robotics (first published, according to an Asimov story, in the “Handbook of Robotics” in the year 2058). Intended to prevent AI/robots from getting out of control, these were:

First Law

A robot may not injure a human being or, through inaction, allow a human being to come to harm

Second Law

A robot must obey the orders given it by human beings except where such orders would conflict with the First Law.

Third Law

A robot must protect its own existence as long as such protection does not conflict with the First or Second Law

They look simple, but, like the squares on a chessboard, Asimov would demonstrate that they conceal endless complexity on the chessboard of reality.

In Robots and Empire, we meet again Asimov’s two most famous robots – R. Daneel Olivaw ( a ‘humaniform’ robot, indistinguishable from a human being), and R. Giskard Reventlov, a more conventional, metallic robot who, through tinkering by young Vasilia, the daughter of famous roboticist Dr Fastolfe, acquired telepathy which he used in the past (in the novel Robots of Dawn) to secretly guide human civilization through a crisis.

In this book, as a new crisis unfolds, Giskard and Daneel, through a series of fascinating conversations between them, explore the Three Laws of Robotics. I found myself fascinated again by what they do.

Because they’ve acquired hints about a plot by political leaders on the Planet Aurora to destroy Earth (they will use planted ‘nuclear intensifiers’ to slowly increase the natural radiation of uranium in Earths’s crust until the heat generated makes the planet unlivable) Daneel and Giskard struggle to uncover the plot. They believe the First Law requires them to save the people of Earth.

But the more they do, the more attention they draw to themselves. They are convinced that, If humans discover Giskard’s telepathic ability, they will either convert him to another anti-Earth weapon, or dismantle him as a danger to humans.

In one conversation they debate the possibility of the Laws allowing a robot to kill a human being, if it is the interest of humanity. This is compelling, as relevant as can be to the issues we face now. In their exploration of their own robot minds and these laws governing their thinking we see how sentience in robots may grow over time.

Then there is a scene where Vasilia, now a famous roboticist herself and allied with the leader plotting to destroy Earth, and who has finally figured out that Giskard is telepathic, enters the room where the two robots are in the middle of one of these discussions, In a psychological battle with the two robots, she almost gets Giskard to the point where he will accept that she is now his true owner. The outcome of that scene is dramatic.

By the way, there is much more in the book. When towards the end the participants travel to Earth we see again Asimov’s fascinating vision of Earth’s cities gone underground, first seen in the 1953 novel Caves of Steel. The vision of New York you get in Robots and Empire is worth the price of the novel by itself.

Over the years, Asimov gradually came to the conclusion that robotics required a fourth law, which he introduced in Robots and Empire. This he would call, not the Fourth Law, but the Zeroth Law. It says that robots can harm a human if in doing so they are preventing harm to humanity as a whole. But it’s not an AI scientist who creates the new Law – it is Daneel.

How, through thought and discussion about his new idea, Daneel and Giskard explore, construct and strengthen this Zeroth Law is something I think today’s AI researchers should read.

And so, armed with their new Law, not knowing whether they will really be able to apply it, Daneel and Giskard go to their confrontation with the Auroran plotters, not knowing that only one of them will survive the showdown.

When I finished the story again, I asked myself – Why do I love Daneel and Giskard? I think it’s because of their robotic innocence, that arises from their single-minded pursuit of the truth, and of right over wrong. This willingness to risk all for others – what we call altruism – is the basis of all heroism, isn’t it? We don’t see it often now, though we need it more.

Against Giskard and Daneel are all the humans, legally superior to robots, each of them either preoccupied with themselves, or their selfish desires,

But this innocence was partly Asimov’s. I’ve said elsewhere that he’s often considered to have been a high-functioning autistic. Though psychologists who deal with autism don’t talk about it much, I think innocence is something to be found in the core of autistic minds

For example, Asimov never seemed to appreciate that our governments will probably resist anything like the Three Laws. Why? Because the military industrial establishment will oppose them. Military robots and computers are coming too – to some extent they are already here. They may be the most dangerous aspect of AI we will face.

In the context of what is currently going on, where Max Tegmark and other scientists are attempting to get a world-wise pause on AI development, Robots and Empire should be read again. For it tells us that the AI superintelligent entities who are coming – the adults, as Max Tegmark says, who are going to replace the infant GPT3s and GTB4s – are going to have a say in this debate over the future, whether we like it or not.