By Rafael Laguna de la Vera and Thomas Ramge*
In Japan, intelligent robots are viewed as friends. In the USA and China, they are seen as servants. And in Europe, they are often perceived as enemies. Culture shapes how we use, build, and regulate technology. But technology is a global phenomenon. So maybe we are better served if we all look at robots with a new, common perspective — what if we perceived them as if they were children? For children to turn out well, parents must raise them responsibly. We humans brought these “robot kids” into the world. At present, many of their skills are less advanced than those of toddlers. However, our artificial offspring possess immense potential. They could grow up to be scientists and economists, doctors and nurses, bus drivers and pilots, architects and artists. Just as parents want their biological children to realise their potential and become contributing members of society, so too should we aspire for our robotic progeny.
In our culture, a humanist worldview serves as an excellent foundation for achieving this. Education is most successful when it enables young adults to navigate the world with self-confidence and self-determination. They should be equipped with a compass of humanist values and the capabilities to positively impact their family, friends, work environment, and society at large. These should be our goals when educating robot children as well.
What is particularly appealing about this approach is that educating intelligent machines is often easier than dealing with stubborn children. If the machines perform well, they can be duplicated. If not, they can simply be shut down. We can design and train them to fulfil our specific needs and values. Nonetheless, this is a hefty responsibility. As AI systems become increasingly autonomous in interacting with the human world, they must develop the ability to discern appropriate from inappropriate behavior on their own.
In familiar scenarios, this is already relatively straightforward. Yet these “robot children” only become genuinely intelligent when they can make informed decisions in unfamiliar situations based on a set of values. It will not be enough to program them with a catalog of rules, like an extended version of Asimov’s Laws. Future generations of AI will need a comprehensive worldview and a broad base of knowledge comparable to what we teach our biological children. This includes recognising their limitations and knowing when to seek help.
For their part, machines are already better learners than humans in some respects. When one robot errs, its “siblings” can learn from that mistake in the next software update. Furthermore, it is crucial to consider the potential long-term consequences. While it still sounds like science fiction, should a powerful AI ever emerge that becomes independent of human control, we would do well to have imbued it with a humanistic worldview beforehand. This way, we aim to prevent dystopian outcomes before they happen.
Admittedly, the metaphor of "robot children" is imprecise and has its limitations. However, this comparison could be apt and useful when considering their development, education, and the roles we envision for them. After all, we are all products of our parents, schools, and society. Who knows? We might even fare better with our robot "offspring."
--
*The article first appeared @ tech.eu's summit website. You can read the original: here
***Rafael Laguna de la Vera is the Director of the Federal Agency for Disruptive Innovation and the founder of several software companies. Thomas Ramge is a non-fiction author and an associated researcher at the Einstein Center Digital Future. In October, MIT Press will publish Laguna de la Vera and Ramge's book On the Brink of Utopia. Reinventing Innovation to Solve the World’s Largest Problems.
