LAST week, regulators in the US announced plans to review export controls across a wide range of emerging technologies. The list includes artificial intelligence and machine learning, as well as technologies that would make substantial use of AI and machine learning, such as robotics and brain-computer interfaces, and supporting technologies. This may result in the US becoming the first nation to explicitly control the spread of AI technologies.
Law-makers are responsible for balancing economic prosperity and growth against public safety threats. Companies such as Google, Apple, Uber and Salesforce have invested heavily in products and services that use AI and machine learning, granting the US a competitive advantage.
Advertisement
But there is a growing concern about intellectual property theft by foreign countries, which corrodes the competitive advantage of these companies. Export controls would make it harder for foreign agents to obtain and reverse engineer US AI and machine-learning products. In addition, these technologies have military applications, such as the design of autonomous weapons and cybersecurity tools.
Preventing the spread of the mathematics underlying AI is impossible. There are broad categories of well-known AI algorithms, including reinforcement learning and recurrent neural networks. The maths is taught in universities and discussed by researchers in scientific literature. It is extremely unlikely that there is a broad class of algorithms that we do not yet know about, kept under wraps by a company or government.
While there might not be any secret algorithms, there are engineering secrets. The process of creating a product or service from an algorithm requires a multitude of design choices that don鈥檛 change the algorithm itself, such as programming language, how data checks are applied and how data is moved from computer memory to processor.
There are also secret data sets that don鈥檛 change the algorithm, but affect the accuracy of its outputs. These software and hardware details can be critical to the commercial success of a product or service. There is a valid argument that these engineering details should be protected, and they already are by national and international rules on intellectual property and trade secrets.
If implemented poorly, export control of AI and machine-learning technologies could result in government control of AI research, which would have a chilling effect on the work done, and a detrimental impact on US economic competitiveness. To protect this industry, we must have clarity about what is to be controlled, and why existing rules and laws are insufficient.
This article appeared in print under the headline 鈥淩obots without borders鈥�
