The Mobileye acquisition is proof that AI needs guidelines

AI, machine learning

GUEST:

The sale of Mobileye to Intel is not surprising when you consider that last year MIT Technology Review ranked the Jerusalem-based company as one of the 10 smartest companies in the world.

Also not surprising is the joy, bordering on hysteria, being displayed by the Israeli government, which has declared its intent to cut taxes as a result of the deal (though it has not been clear exactly how). In addition, the government has used the occasion of Mobileye’s purchase by Intel as an opportunity to tout its success in the area of innovation.

However, the story here is bigger than Mobileye.

The real story is the relationship between government, society, and technology.

The challenge presented by artificial intelligence (AI) is to get machines to solve problems that are uniquely the province of humankind: use of language, development of abstract thoughts and concepts, and, of course, self-improvement. This brave new world began with such technological leaps forward as personal assistants, continued with fundamental changes to the medical and legal systems, and is today on the verge of creating autonomous cars and weapons.

Technological breakthroughs, along with major investments in research and the timely acquisition of companies, are bringing us closer to the day when these challenges will be addressed. It is likely that within a few short years, the AI revolution will enhance our lives and streamline the systems we’ve come to rely on.

However, this revolution is not without challenges.

Decision-makers tend to vacillate between an irrational fear of artificial intelligence — “What will happen when robots take over our lives?” or “What will happen once unmanned vehicles must decide whether to kill the person sitting inside it or that child who just ran out to the middle of the street?” — and euphoria and technological determinism, based on the belief that “you cannot stop technology.” These extreme reactions all lead to psychological discomfort that makes it difficult to deal with the real issues at hand.

Based on the lessons of the past, we can learn that new technological realities, particularly with regard to the revolutions in information and connectivity of the past two decades, tend to be recognized too late by policy makers. Usually their reactions are heavy on euphoria but desperately lacking in basic digital literacy. One telling local example comes to mind, for me: The Israeli Ministry of Communications, preoccupied with taking over public broadcasting in Israel, has for years put off developing a set of standards for the Internet of Things.

To avoid similar problems, the time has come to develop a set of guidelines to address the challenges presented by AI. This proposal can be categorized into four main areas of focus.

First, we must tackle the broad policy aspects and potential impact of AI on the economy, society, and government, from the future of transportation to the fate of democratic forms of government in an era of tailor-made, accurate forecasting systems.

Second, a paradigm shift regarding state regulation over matters related to machine learning must take place to ensure competition and deal with licensing, taxation, cybersecurity, and quality assurance.

Third, we’ll need to protect human rights and create new standards in administrative law for smart systems, whether deployed as an effective and inexpensive substitute for the judgment of officials working in state bureaucracies or as a way to monitor systems in the private sector.

Drilling down further on this particular focus, the autonomous systems being developed by Mobileye are based on the concept of reinforcement learning, whereby machines “learn” through experience after repeated exposure to driving videos. In this way, the system learns how to respond independently to situations that it has not previously experienced. With reinforcement learning being hailed as the technological breakthrough of 2017, we must develop ways to ensure systemic transparency, since even innovative programming systems’ developers cannot possibly know to where independent learning will lead.

Fourth and finally, a comprehensive action plan to deal with such technological advances must be developed at the institutional and organizational levels within government. Experts who understand and can analyze the interactions between artificial intelligence technologies and social values will be necessary. And a cross-governmental dialogue on the subject must be initiated.

While the blockbuster Mobileye deal is (and should be) a source of pride for all of Israel, in that it will make Israel the central address for automatic cars in the world and reinforces the perception that Israel really is the startup nation, the current challenge is to turn Israel into a startup society, too.

The Mobileye deal proves that the future is now.

Dr. Tehilla Shwartz Altshuler is director of the Government in the Information Age program at the Israel Democracy Institute.

Machine Intelligence Landscape

Above: The Machine Intelligence Landscape This article is part of our Artificial Intelligence series. You can download a high-resolution version of the landscape featuring 288 companies by clicking the image.

We will be happy to hear your thoughts

Leave a reply