Tools and Weapons: The Promise and the Peril of the Digital Age

by Brad Smith

Einstein’s words speak to the crux of today’s challenge. As technology continues to advance, can the world control the future it is creating?

control,” he said. “We had a police station at the school and there were cameras in every corridor. Everything was tracked from grades, SAT scores, attendance, and little stickers on our student IDs that allowed us to use the internet.”

Schrems proudly recalled how he helped his American peers circumvent the blocks the school had put on Google searches. “I showed them that there’s Google.it, which works perfectly fine because the school only blocked dot-com,” he said. “The exchange student introduced the school to international top-level domains!” It was a relief to return to Vienna, he told us, “where we have so much freedom.”

It concluded that Europe’s national data protection authorities are empowered to make their own assessments of data transfers under the agreement. In effect, the court gave more authority to independent regulators it knew would be tougher in reviewing privacy practices in the United States. Immediately people wondered

Even the ultimate implications for China are weighty. Over time, the European approach can lead to mounting pressure on China to confront an important crossroads. It can move forward without privacy protection for data within its borders, or it can strengthen its economic connections with Europe with the inevitable data flows this will require. But it will become more difficult to do both.

Like the reaction to many near-disasters, however, the immediate response to the negotiation of the Privacy Shield was mostly a sigh of relief. It was another wake-up call, but again people hit the snooze button. Data flows could continue, and companies could continue to do business. Most tech companies and government officials postponed deeper thinking about the longer-term geopolitical implications for another day.

For a diversified tech company like Microsoft, the impact of the GDPR could hardly have been more intense. We had more than two hundred products and services, and many of our engineering teams had been empowered to create and manage their own back-end data infrastructures. There were certain similarities, but there were also important differences in the information architecture used in different parts of the company.

In early 2016, we assembled a team with some of our best software architects. They had two years before the GDPR would take effect on May 25, 2018, but they had no time to spare. The architects needed first to turn to lawyers, who defined what the GDPR required. With the lawyers, they then created a specification that listed all the technology features our services would need to enable. The architects then crafted a new blueprint for the processing and storage of information that would apply to all our services and make these features effective.

The engineering and legal teams walked through the blueprint, timelines, and resource allocations. It impressed everyone. And in some ways, it surprised everyone. As the meeting progressed, Satya suddenly exclaimed with a bit of a chuckle, “Isn’t this great?” He continued, “For years it has been next to impossible to get all the engineers across the company to agree on a single privacy architecture. Now the regulators and lawyers have told us what to do. The job of creating a single architecture just got a whole lot easier.”

Given the large, diversified, and empowered engineering structure at Microsoft, this challenge was sometimes greater than at other tech companies. It had led us in the past sometimes to maintain for years two or more overlapping services, an approach that almost never turned out well. Apple, in contrast, had sometimes relied on its narrower product focus and Steve Jobs’s centralized decision making to solve this problem. It was perhaps ironic, but European regulators were doing us something of a favor by defining a singular approach that required engineering compromise all around.

Satya signed off on the plan. Then he turned to everyone and added a new requirement. “As long as we’re going to spend all the time and money to make these changes, I want to do this for more than ourselves,” he said. “I want every new feature that’s available for our use as a first party to be available for our customers to use as a third party.” In other words, create technology that could be used by every customer to comply with the GDPR. Especially in a data-dominant world, it made complete sense. But it also added more work. All the engineers in the room gulped. They left the meeting knowing they’d need to put even more people on the project.

Others in the tech sector nonetheless often pointed to the American public’s ambivalence toward privacy as reason to ignore US regulatory pressures. “Privacy is dead,” they’d say. “People just need to get over it.”

As Mactaggart quizzed one of his guests about his job at Google, he wasn’t just dissatisfied with the answers, he found them terrifying.

What private data were tech companies collecting? What were they doing with it? And how can I opt out? If people knew what Google knew, the engineer replied, “They’d freak out.”

The Americans stranded in the dial-up era aren’t confined to Ferry County. They are in every single state in the country. According to the FCC’s 2018 broadband report, more than twenty-four million Americans, more than nineteen million of whom live in rural communities, lacked access to fixed high-speed broadband.2 That’s roughly the population of New York state.

The highest unemployment rates in the country are frequently located in the counties with the lowest availability of broadband, highlighting the strong link between broadband availability and economic growth.10

In a world reliant on modern high-speed access to data, an area without broadband is a communications desert.

Rural counties across the country like Ferry County had helped put a populist into the White House. We had started our trip in King County, where Seattle is located and where only 22 percent voted for Donald Trump. In Ferry County, only 30 percent voted for Hillary Clinton.11

People sometimes look at us quizzically when they hear that the Airband initiative will reinvest revenue from telecommunications partnerships rather than make a profit. Why would a company spend its money this way? As we point out, the entire tech sector, including Microsoft, will benefit when more people are connected to the cloud. In addition, we’re building new applications

that people can put to work in rural areas once they’re connected. One of our favorites is called FarmBeats, which uses TV white spaces to connect small sensors across farmland to enable precision techniques that improve agricultural productivity and reduce environmental runoff. If we can find new ways to combine doing good with doing well, we open the door to even more investments that can reignite economic growth in rural areas.

Ultimately, we need a national crusade to focus on and close the broadband gap. We need to recognize that, as was the case with electricity, a country separated by broadband availability will

remain a nation more divided overall.

But in a world where today’s hits quickly become yesterday’s memories, a tech company is only as good as its next product.

And its next product will only be as good as the people who make it. This means, in short, that technology is fundamentally a people business.

The implications are multifaceted and even profound. To succeed in the digital era, companies need to recruit world-class talent, both homegrown and from elsewhere. Local communities need to ensure their citizens are equipped with new technology skills. Countries need immigration policies that give them access to the world’s top talent. Employers need to develop a workforce that reflects and understands the diversity of the customers and citizens they serve. This requires not only bringing more diverse people together but also creating a culture and the processes that will enable employees constantly to learn from each other. Finally, as technology accelerates growth in key urban centers, these regions need to manage the challenges this growth is creating, not just for individual institutions, but for the entire community.

MSR is one of the world’s largest organizations dedicated to basic research. It’s hardly typical, as it reflects an elite of the elite when it comes to people creating technology. But it provides an important initial window on the world of technology. MSR has more than twelve hundred PhDs, eight hundred of whom have computer science degrees. To put that in perspective, the computer science departments of major universities typically employ sixty to a hundred PhDs as faculty and postdoctoral fellows. And in quality, MSR typically is considered a match for any of the top universities. Think of it as one of the world’s best university computer science departments multiplied by ten in size. It’s a modern-day counterpart to what AT&T assembled at Bell Labs decades ago.1

There were two Americans and one person each from Finland, Israel, Armenia, India, Iran, and China. All eight now lived in the Seattle area and worked together on our Redmond campus. This group of researchers personified something much larger than itself. Here was a team working on one of today’s great technology challenges, which required a world champion lineup—one that America’s immigration system had enabled us to bring together.

Its open door to German rocket scientists after World War II was critical to sending the first man to the moon. With the help of federal investments in basic research at the country’s great universities and President Eisenhower’s support of math and sciences in the nation’s public schools,3 the United States developed an approach to research, education, and immigration that led to decades of global economic and intellectual leadership.

The rest of the world studied this model and increasingly emulated it. But Americans increasingly forgot what made it work. And the political support for its various pieces began to fall apart.

Microsoft Philanthropies made the new program—Technology Education and Literacy in Schools, or TEALS—a cornerstone of its educational mission. It annually enlists 1,450 volunteers from Microsoft and 500 other companies and organizations to teach computer science in almost 500 high schools in 27 US states, plus the District of Columbia and British Columbia, Canada.

While you would be hard-pressed to say that every student must take computer science, you could say that every student deserves the opportunity.

When executive director Naria Santa Lucia joined the effort five years ago, she focused on providing students with mentors, internships, and connections to potential employers. This has created roles for businesses and individuals across the community. The combination has not only led to strong graduation rates for the students, it has provided a clear path into well-paying jobs.

In short, the students taking the AP computer science class are more male, more white, more affluent, and more urban than the country as a whole. This part of the problem has multiple causes. But the tech sector needs to accept its share of responsibility. It has not always been an easy place for women or minorities to build a career. Science and technology have long had prominent female pioneers, including Marie Curie, who remains the only person to twice win the same Nobel Prize, and Bertha Benz, the first person to show the world the automobile’s potential.31 But while men were prepared to acknowledge the contributions of these women as individuals, the world of technology remained stubbornly slow in recognizing and creating opportunities for women more broadly. At most tech companies, women still represent less than 30 percent of the workforce, and an even lower percentage of technical roles. Similarly, African Americans, Hispanics, and Latinos typically account for less than half of what one would expect based on their representation in the American population.

As companies like Amazon and Microsoft continued to grow, we were joined by burgeoning engineering outposts from more than eighty companies based in the Silicon Valley. Suddenly the Seattle area had evolved from the Emerald City to Cloud City. Between 2011 and 2018, median home prices increased by 96 percent, while median household income rose by only 34 percent.32

We shared the skepticism about taxing jobs but felt that the business community needed to do more than criticize the measure. It needed to step up in a bigger way. Seattle’s mayor and city council canceled the head tax, but little emerged in terms of an effective alternative.35

The gap was even bigger in the smaller cities outside Seattle, where construction of both low- and middle-income housing had stagnated. People in low- and middle-income families increasingly were being forced into towns and suburbs much farther from their jobs. The region now ranked among the worst in the country for the percentage of people enduring daily commutes in excess of ninety minutes.37

All these challenges require action. As I sometimes say at Microsoft when we start a new project, first prize is to do something big. Second prize is to do something. Success rarely comes to people who do nothing.

The challenge was that the tech sector, to its credit, always looked forward. The problem was that, to its detriment, too few people spent time or even accepted the virtue of looking in the rearview mirror long enough to use a knowledge of the past to anticipate the problems around the corner.

Computers were becoming endowed with the ability to learn and make decisions, increasingly free from human intervention. But how would they make these decisions? Would they reflect the best of humanity? Or something much less inspiring? It had become increasingly apparent that AI technologies desperately needed to be guided by strong ethical principles if they were to serve society well.

“AI is a computer system that can learn from experience by discerning patterns in data fed to it and thereby make decisions.”

Three recent technological advances have provided the launchpad that has let AI take flight. First, computing power finally advanced to the level needed to perform the massive number of calculations needed. Second, cloud computing made large amounts of this power and storage capacity available to people and organizations without the need to make large capital investments in massive amounts of hardware. And finally, the explosion of digital data made it possible to build massively larger data sets to train AI-based systems. Without these building blocks, it’s doubtful that AI would have accelerated so quickly.

But it took a fourth foundational element, which has been critical in helping computer and data scientists to make artificial intelligence effective. This goes to the second and even more fundamental technological capability needed for AI, cognition—in other words, the ability of a computer to reason and learn.

In many ways, we go about our human lives not by reasoning through rules, but by recognizing patterns based on experience.

This approach uses statistical methods for pattern recognition, prediction, and reasoning, in effect building systems through algorithms that learn from data.

During the last decade, leaps in computer and data science have led to the expanded use of so-called deep learning, or with neural networks.

Our human brains contain neurons with synaptic connections that make possible our ability to discern patterns in the world around us.7 Computer-based neural networks contain computational units referred to as neurons, and they’re connected artificially so that AI systems can reason.8 In essence, the deep learning approach feeds huge amounts of relevant data to train a computer to recognize a pattern, using many layers of these artificial neurons. It’s a process that is both computationally and data intensive, which is why progress required the other advances previously noted....

While each of these four principles was important, we realized that they all rested on two other principles that were foundational to the others’ success. The first was transparency. To us, this meant ensuring that information about how AI systems were making consequential decisions was public and understandable. After all, how can the public have confidence in AI and how can future regulators evaluate its adherence to the first four principles if the inner workings of AI are kept in a black box? Some argue that AI developers should publish the algorithms they use, but our own conclusion was that these were likely to be less than enlightening in most circumstances or reveal valuable trade secrets that would undermine competition in the tech sector. We had already been working in the Partnership on AI with academics and other tech companies to develop better approaches. The emerging focus is on making AI explainable, like describing the key elements that are being used to make decisions.

The final ethical principle of AI would be the bedrock for everything else: accountability. Will the world create a future in which computers remain accountable to people and in which the people who design these machines remain accountable to everyone else? This may be one of the defining questions of our generation.

At the same time, we recognized that some of our own employees were uncomfortable working on defense contracts for the US or other military organizations. Some were citizens of other countries, some had different ethical views or were pacifists, and some simply wanted to devote their energy to alternative applications for technology. We respected these views, and we were quick to say that we would work to enable such individuals to work on other projects. Given Microsoft’s size and diverse technology portfolio, we felt that we could most likely accommodate these requests.

We rolled up our sleeves to learn more and develop more refined views. This led us back to our six ethical principles that informed the ethical issues applicable to AI and weapons. We concluded that three were most at stake—reliability and safety, transparency, and most importantly, accountability. Only by addressing all three of these could anyone maintain public confidence that AI would be deployed in a way that would keep human beings in control.