Let's face it: the pace of progress in artificial intelligence can sometimes seem unsettling. Terminator-style machines remain science fiction, but AI could have a huge impact on employment, introduce bias into algorithms, and contribute to the development of autonomous weapons. But perhaps the biggest looming threat may be making sure we understand how these increasingly complex systems work when they go awry.
The latest evidence that even the experts are concerned about this is the creation of a new AI ethics research center at Carnegie Mellon University. The new center, called K&L Gates Endowment for Ethics and Computational Technologies, is funded with $10 million from K&L Gates, an international law firm based in Pittsburgh.
Anxiety over machine intelligence has been gaining momentum. Last month the White House released a report assessing the potential effects of AI. And several of the world's largest tech companies recently joined forces to create an organization, called Partnership on AI, to study the technology and its potential impacts.
In a statement, CMU's president, Subra Suresh, said it will be important to consider the human side of all AI systems. “It is not just technology that will determine how this century unfolds," he said. "Our future will also be influenced strongly by how humans interact with technology, how we foresee and respond to the unintended consequences of our work, and how we ensure that technology is used to benefit humanity, individually and as a society."
CMU itself is experiencing some teething pains due to advances in AI. Last year its robotics department was raided by Uber for a nearby research center dedicated to automated driving. At the same time, the university is spinning out AI-powered startups and consulting with big companies on various AI projects.
Besides unemployment, algorithmic bias, and autonomous weapons, one of the most significant—and least appreciated—consequences of AI could be the way we come to rely on systems that are inscrutable because no one programmed them. This issue is already appearing in some situations, while some experts are trying to devise machine-learning systems that are able to explain their workings.