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

by Brad Smith

“Almost no one around here has broadband,” he said. “Promises have been made for years, but nothing’s been done.” But according to data from the Federal Communications Commission, or FCC, every person in Ferry County had access to broadband.

evidence that the percentage of Americans without broadband access is much higher than the FCC’s numbers indicate. As we analyzed this data, we found that it was based on flawed methodology. The FCC concludes that a person has access to broadband if a local service provider reports that it could provide such service “without an extraordinary commitment of resources.”5 But many such companies don’t provide this service in practice. It’s like telling someone they have access to a free lunch if a local restaurant says it could serve it to them if it wanted.

When you talk to business leaders about where they might expand their operations and add jobs, this requirement surfaces almost immediately. Asking them to open a new facility in a place without broadband is like asking them to set up shop in the center of the Mojave Desert. In a world reliant on modern high-speed access to data, an area without broadband is a communications desert.

Center for Rural Affairs also has details and numbers to make the economic case for broadband adoption. Its 2018 report titled Map to Prosperity shows that eighty new jobs are created for every one thousand new broadband subscribers.13 An increase of four megabits per second in residential broadband speed translates to an annual increase in household income of twenty-one hundred dollars. And people looking for work find a job 25 percent more quickly through online searches than through more traditional approaches.

It took forty years for the landline telephone to reach 90 percent penetration, yet the cell phone reached the same threshold in just a decade. You never hear about the world needing to solve a radio or TV gap—these wireless devices were adopted quickly and were plug-and-play, latching on to the right frequency to work.19 The lesson is obvious: If it’s possible to shift from fiber-optic cables to wireless technology for broadband, we can spread broadband coverage farther and faster and at a lower cost—not just in the United States, but around the world.

It was in 2011 at a United Nations conference in Nairobi, Kenya, and we enabled attendees to use the Xbox at broadband speeds over the Internet based on a TV white spaces signal that traveled a mile.

By 2017, we concluded that TV white spaces technology was ready for broader adoption at scale, including in rural areas across the United States. After several months of planning, we launched in July what we called the Microsoft Rural Airband Initiative at the Willard InterContinental hotel in Washington, DC. We pledged to bring broadband coverage to two million additional Americans in rural areas within five years—by July 4, 2022. We would not enter the telecommunications business, but we would partner with telecom providers and deploy a mix of wireless technologies, including new wireless devices using the TV white spaces spectrum. We pledged that for five years we would reinvest every dollar of profit from these endeavors to further expand coverage.

One key to deploying this strategy is to use the right technology in the right places. We expect that TV white spaces and other fixed wireless technologies will ultimately provide the best approach to reach approximately 80 percent of the underserved rural population, particularly in areas with a population density between two and two hundred people per square mile.

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.

90 percent of city dwellers had electricity versus 10 percent of rural Americans24—a gap not seen in other Western countries. At the time, electricity powered homes and barns in almost 95 percent of the French countryside.25

When President Roosevelt died in 1945, nine out of ten farms in rural America had electricity.32 Through public-private partnerships, persistence, and a little ingenuity, the United States had managed to shrink the rural electricity gap by 80 percent in ten years—all during an arduous economic recovery and the Second World War.

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.

In each of these areas, tech companies are dependent on support from a community and often even a nation. And in each area, tech companies have an opportunity and a responsibility to do more themselves. It’s a formidable challenge, much like a Rubik’s Cube puzzle that can only be solved by moving many pieces at the same time.

Microsoft Research, or MSR, as everyone calls it. 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 is simply no way the United States would be the global leader in information technology if it had not attracted many of the best and brightest people in the world to come work at leading universities or live in technology centers around the country.

Immigration’s role in innovation was important to the United States when the country’s West Coast economy was still dominated by agriculture, and silicon was associated only with sand. The country’s ability to attract Albert Einstein from Germany at the height of the Great Depression played a vital role in awakening President Franklin Roosevelt to the need to create the Manhattan Project.2 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.

Away from the public glare of politics, we sometimes encounter a similar situation. A tug-of-war emerges over a single issue in the world of business or regulation. It becomes a contest that inevitably will produce one winner and one loser. It’s a recipe for a sustained impasse, for getting nothing done. Ironically, the answer to such problems is sometimes to broaden the challenge. One tenet I always employed in negotiations was a simple one: Never let a negotiation narrow to a single issue that can produce only one winner, even if it means holding open some other topics on which agreement might seem in reach. Instead, broaden the discussion so that more issues are on the table. Create the opportunity for more give-and-take and a round of compromises that can produce a scenario that enables everyone to claim victory in the final stage. It was an approach that had proven indispensable

He paced around my office, as he often did. He then sat down, but not in a chair. Instead, he sat on top of my desk cross-legged—something I hadn’t seen before. He looked me straight in the eye and said, “If we’re going to spend all of this money, I want you to ensure that some real people get some real benefit from this.” I promised that I would.

But there’s a huge shortage of teachers who can teach computer science. Less than 20 percent of the nation’s high schools offer the Advanced Placement course in the field.12 The number of young people taking the AP course in 2017 was lower than for fifteen other subjects, including European history. One challenge is the high cost of training teachers so they can teach computer science.

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.

Singapore’s lifelong learning financial accounts

We concluded that if we were going to put the company’s reputation on the line advocating for an additional tax on local businesses, we should propose a structure that would require Microsoft and Amazon, as the two largest tech companies in the state, to pay a higher tax rate than everyone else.

While the proposal initially raised some friction with other companies,25 legislators found their way to a compromise that capped the new tax surcharge at $7 million per company per year. Just six weeks later, the legislature approved a new budget with a dedicated fund that will raise roughly $250 million per year for higher education. Washington’s new Workforce Education Investment Act was hailed both locally and nationally for its commitment to “provide free or reduced tuition for lower- and middle-income students attending community colleges and public institutions, provide new funding for strapped community colleges and eliminate wait lists for financial aid beginning in 2020.”26 As described by a professor at Temple University, it was “pretty much the most progressive state higher ed funding bill” enacted in several years.27 For me, it was proof that if the tech sector could get comfortable with a more community-focused approach and pay perhaps even a bit more than its share toward it, we could have a real and positive impact.

Unfortunately, this type of progress remains a bit of an oasis in the desert. In the United States, access to technology skills is far from distributed evenly. Like the broadband gap, the skills gap hits some groups far harder than others. It’s exacerbating almost every other divide that afflicts America.

You see the disparate impact clearly when you look at the students learning computer science. At a time when technology suffers from a shortage of women, only 28 percent of the students taking an AP computer science exam in 2018 were girls.28 You see the same trend play out with underrepresented racial minorities. These groups comprised just 21 percent of the students taking these exams, compared to their 43 percent representation in the country.29 And at a time when the country worries about economic opportun...

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 n...

Most weekday afternoons in Silicon Valley, the highway feels like a parking lot—except that you can drive a bit more quickly through a parking lot. The strains on highways are like the visible part of the iceberg. It’s the easiest to see, but growth places the same demands on every part of a region’s infrastructure, from transit systems to schools. Over the past few years, the problem has reached a much deeper level. While jobs have grown, the housing supply has not kept pace. Basic economics goes to work.

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

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

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.

industry, like most, often rushes forward with an innovation without helping people understand fully what it is or how it works. This is coupled with what for too long was an almost theological belief that new technology will be entirely beneficial. Many in Silicon Valley long believed that government regulators could not keep up with technology.

Even the best technologies have unintended consequences, and the benefits are seldom spread uniformly. And this is before the new technology is misused for harmful ends, as it inevitably will be. In the 1700s, soon after Ben Franklin created the postal service in the United States, criminals invented mail fraud. In the 1800s, with the telegraph and the telephone, criminals invented wire fraud. In the twentieth century, when technologists invented the internet, it was apparent to anyone who knew history that the invention of new forms of fraud was unavoidable.

There is no universally agreed-upon definition of AI across the tech sector, and technologists understandably advance their own perspective with vigor.

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

When people a hundred years from now look back at the history of the twenty-first century, they’ll likely conclude that the decade from 2010 to 2020 was when AI came together.

One approach was based on what’s called “expert systems.” Especially popular in the late 1970s and 1980s, this involved the collection of large amounts of facts and the creation of rules that computers could apply in chains of logical reasoning to make decisions. As one technologist has noted, this rule-based approach couldn’t scale to match the complexity of real-world problems. “In complex domains, the number of rules can be enormous, and as new facts are added by hand, keeping track of exceptions to and interactions with other rules becomes impractical.”5

Since the 1980s, an alternative approach to AI has proved superior. This approach uses statistical methods for pattern recognition, prediction, and reasoning, in effect building systems through algorithms that learn from data.

Those more immediate issues had come up that same year at a conference sponsored by the White House. The talk of the conference had been an article published in ProPublica titled “Machine Bias.”13 The subtitle for the piece said it all: “There’s software used across the country to predict future criminals. And it’s biased against blacks.” AI was increasingly being used to make predictions in a wide variety of settings, and concern was mounting over whether the systems were biased against certain groups in various scenarios, including people of color.14

The bias problem described by ProPublica in 2016 was a real one. It reflected two real-world causes that each need to be addressed if AI is to perform the way the public rightfully expects. The first involves work that includes biased data sets.

MIT researcher Joy Buolamwini, a Rhodes scholar and poet, and Stanford University researcher Timnet Gebru undertook work to advance public understanding of AI bias by comparing the facial-recognition accuracy rates for people of different genders and races. The two women documented higher error rates, for example, in identifying gender for the faces of black politicians in Africa compared to white politicians in northern Europe. As an African American woman, Buolamwini even found that some systems identified her as a man. Buolamwini and Gebru’s work helped reveal a second dimension of bias that we also need to consider. It’s difficult to build technology that serves the world without first building a team that reflects the diversity of the world. As they found, a more diverse group of researchers and engineers is more likely to recognize and think harder about bias problems that may even impact them personally.

As computers gained the ability to make decisions previously reserved for humans, virtually every ethical question for humanity was becoming an ethical question for computing.

one can’t graduate from an American military academy without taking a course in ethics.29 The same is not yet true for computer science majors at many American universities.

“Could we see a Hippocratic oath for coders like we have for doctors?”

In the tech sector, some employees wanted to play an active role in shaping their companies’ decisions and engagement on the issues of the day. Perhaps not surprisingly, this view was more pronounced at a time when people had less trust in governments. Employees were looking to another institution they hoped might do the right thing and have some influence on public outcomes.

They explained that the Customs and Border Protection personnel used a computer system with a drop-down menu when people initially crossed the border. Agents would classify someone either as an unaccompanied minor, an individual adult, or an adult with children, meaning a family unit. When children subsequently were separated from their parents, the computer system’s design forced agents to go back and change this designation, for example by inputting a child’s name as an unaccompanied minor and the parent’s name as an individual adult. Critically, this overwrote the prior data, meaning the system no longer retained the family designation that previously had listed everyone together. As a result, the government no longer had any record that connected family members. This was not only a story about immigration and families. It was also a story about technology. The government was using a structured database that worked for one process but not for another. Rather than update the IT system to support the new steps involved in separating families, the administration had plunged ahead without thinking about the computer architecture that would be needed. Having seen CBP’s systems at a command center near the Mexican border on a visit with Wendy just months before, I was not surprised that its systems were antiquated. But I was still horrified that the administration had failed to think about the implications of what it needed in terms of basic technology infrastructure.

As we thought it through, we concluded that this new technology should be governed by new laws and regulations. It’s the only way to protect the public’s need for privacy and address risks of bias and discrimination while enabling innovation to continue. To many, it was odd for a company to call on the government to regulate its products.

Our interest in regulation came from our emerging sense of where the market was heading. A few months earlier, one of our sales teams had wanted to sell an AI solution that included facial-recognition services to the government of a country that lacked an independent judiciary and had a less than stellar track record for respecting human rights. The government wanted to deploy the service with cameras across its capital city. Our concern was that a government that flouted human rights could use the technology to follow anyone anywhere—or everyone everywhere.

With the advice of our internal AI ethics committee, we decided we would not move forward with the proposed deal. The committee had recommended that we draw a line and refrain from making facial-recognition services available for generalized use in countries that Freedom House, an independent watchdog that tracks freedom and democracy around the world, had concluded were not free. The local team was not happy.

This need for a principled approach was reinforced when a local police force in California contacted us and said they wanted to equip all their cars and body cameras with a capability to take a photo of someone pulled over, even routinely, to see if there was a match against a database of suspects for other crimes. We understood the logic but advised that facial-recognition technology remained too immature to deploy in this type of scenario. Use of this nature, at least in 2018, would result in too many false positives and flag people who had been wrongly identified, especially if they were people of color or women, for whom there remained higher error rates. We turned down the deal and persuaded the police force to forgo facial recognition for this purpose.

The auto industry spent decades in the twentieth century resisting calls for regulation, but today there is broad appreciation of the essential role that laws have played in ensuring ubiquitous seat belts and air bags and greater fuel efficiency. The same is true for air safety, food, and pharmaceuticals.