People could be forgiven for thinking that the Google event in San Francisco last week would be about new devices and the Android mobile operating system’s latest iteration. The annual ritual is, after all, one of the touchstones of the tech industry. But they turned out to be wrong. It wasn’t about Google’s new Pixel phones. Nor was it about the Home smart speaker. It was about the connective tissue binding them—Google Assistant, the company’s artificial intelligence (AI), a beefed up version of a prior attempt, Google Now. And that raises a number of questions that are proliferating increasingly at the intersection of technology and ethics.
AI routines running under the hood of various digital services and products, and on personal computing devices—think Apple’s Siri and Microsoft’s Cortana—have been commonplace for a few years now. But Google Assistant, along with its predecessor and rival Amazon’s Alexa, is taking aim at something larger.
It’s a seductive vision—ubiquitous environmental AI able to respond to voice input anywhere in the house, sophisticated enough to understand natural language and context and carry on a conversation with the user in the process of offering information and performing tasks such as scheduling appointments or locking the door. The hubs currently take the form of small speakers—Google Home and Amazon Echo, powered by Assistant and Alexa, respectively—but this is the embryonic stage. Google’s ambition is to have multiple hubs or portals in the house. As the Internet of Things inches towards realization, every networked component in the environment—from televisions to refrigerators—could be part of this ecosystem. And others will follow suit; rumour has it Apple is working on its own offering.
This is a future pop culture has been preparing us for since Gene Roddenberry first had Captain Kirk talk to the Enterprise’s computer back in the 1960s. That might explain the ease with which consumers are taking to the concept of voice input. Earlier this year, Mary Meeker’s annual “Internet Trends” report showed that Google voice queries have grown 35-fold since 2008 and that Amazon’s Echo was the fastest-selling speaker in 2015.
But consumers’ comfort with this vision of AI-enabled lives masks issues of privacy that warrant careful consideration. For consumer AI to offer the ease of use that will make it attractive, it must offer as close a facsimile of having a conversation with another person as possible. This is immensely difficult; the ability to understand natural language conversation in context rather than predetermined commands is the holy grail. Two components are necessary: sophisticated algorithms and vast amounts of data. And that includes every scrap of personal data possible.
Data privacy concerns used to largely be the preserve of policy wonks and the tinfoil hat crowd until a few years ago. Multiple leaks and revelations, from WikiLeaks to Edward Snowden, changed all that. But the public focus remains for the most part on government snooping. When it comes to private actors, consumers are voting with their wallets; lack of privacy and surrendering personal data is a fair price to pay for more convenient lifestyles.
Is this necessarily a negative? That depends on the best practices tech companies develop—Apple, for instance, is focusing on protecting users’ data in order to differentiate its products—and on the outcome of their ongoing tussles with governments around the world that demand access to user data for legal purposes. It does, however, point to those ethical questions that the rise of AI is throwing up.
The industry is well aware of these questions, and of the need to show that it is taking concerns on board. Thus, Amazon, Facebook, Google’s DeepMind division, IBM and Microsoft founding a new organization called the Partnership on Artificial Intelligence to Benefit People and Society that aims to initiate a wide dialogue about the nature, purpose and consequences of AI. Tesla CEO Elon Musk has also been harping on the issue for a while now; last year, he was among the founders of OpenAI, another organization aimed at addressing such issues.
They have their work cut out. AI ethics have much more to address than just user privacy. Imagine, for instance, a bank using AI to recommend or screen loan applicants, and the algorithm using causal relationships to discriminate on the basis of gender or caste or race. Or, for that matter, the multiple implications of AI deployed in a military context or controlling driverless vehicles—or the issue du jour, employment. These are also eminently plausible scenarios.
It’s an inevitable future; technological genies can never really be put back in the bottle. But the rise of AI cannot be left to the industry; it demands the involvement of everyone from social scientists to ethicists and philosophers. Social and technological paradigm shifts are rarely painless, after all.
Is giving up privacy in exchange for the convenience of AI a fair deal? Tell us at firstname.lastname@example.org