“When you have a compounding problem, you need a compounding solution,” added Hoffman. The jobs issue “is a power law problem, and the only way to solve a power law problem is with a power law solution” for improving humanity’s ability to adapt. Turning more forms of AI into more forms of IA is that solution.
Ma Bell’s Intelligent Assistance
I visited a lot of companies in researching this book, and none was more innovative in creating intelligent assistance to help its employees become lifelong learners than old, reliable AT&T. Don’t let the moniker “Ma Bell” fool you. Don’t let AT&T’s CEO Randall Stephenson’s good ol’ boy Oklahoma accent take you in. Don’t let the mild Midwest demeanor of Bill Blase, their head of human resources, get you to drop your guard. And whatever you do, don’t move your eyes off their chief strategy officer, John Donovan, and Krish Prabhu, the head of AT&T Labs, because they will disrupt your business on behalf of one of your competitors before breakfast. They might even do it just for fun.
Attention, Kmart shoppers: this is not your grandma’s Ma Bell!
Back in 2007, when AT&T found itself pioneering software-enabled networks to handle the explosion of data created by the iPhone, for which it was the original exclusive network provider, it realized that it had to up its own innovation metabolism—broadly and quickly. If you’re running with Apple, you need to be able to run as fast as Apple. In 2016, AT&T was still at it. That year it opened up an “Internet of Things Foundry” in Dallas, an innovation shop full of network engineers. They invited customers in with this proposition, explained Vice Chairman Ralph de la Vega: “Tell us what problem you want us to solve, and we commit that within two weeks we’ll give you a prototype solution for you that works on a real live network … Every time we do this, it results in a contract.”
So, for instance, the global shipping giant Maersk needed a sensor that it could affix to every shipping container it owns, enabling the company to track its containers anywhere in the world. The sensor had to affix to two hundred thousand cargo refrigerator containers, it had to be able to measure their humidity, temperature, and whether they had suffered any damage, and it had to broadcast that data to their headquarters, and—this was the real catch—the sensor had to operate without batteries and be able to last ten years, because they couldn’t be changing them all the time. In two weeks, AT&T engineers built a prototype of a sensor, half the size of a shoe box, that could affix to every Maersk container and was powered by a combination of solar and kinetic energy.
What happened to AT&T was that the supernova transformed its business overnight. Its decision to virtualize networks to expand its capacity made it more of a software/networking company, and then it really struck gold with the rise of big data, which meant that the data and voice traffic AT&T was carrying over its wires could be aggregated and anonymized and then mined for trends. So suddenly, as noted earlier, AT&T, using wireless cell phone data, could tell a signage company how many people who drove by their sign on the freeway ended up shopping in the store advertised on that sign—and if the sign became digital, and changed every hour, they could tell them which message was most effective. AT&T started telling some customers, heck, we’ll cut you a deal on the transmission costs if we can mine the data and use it to solve customer problems or puzzles. In the blink of an eye your friendly phone company became an all-around business solutions company, also competing with IBM or Accenture.
Precisely because Stephenson understood that for his company to thrive it had to be the networking enabler and solutions provider to the most disruptive companies in the world, he knew he had to disrupt his own workforce in the process.
“We felt a fundamental obligation to reskill our workforce,” said Donovan. “We needed a smaller and smarter workforce. STEM skills are now table stakes.” But they also knew that when you’re dealing with three hundred thousand employees, if you want to challenge them to upgrade their skills, you need a strategy of what I call “intelligent assistance” to provide the scaffolding and incentives that make a new learning journey for so many people sustainable.
AT&T’s version of intelligent assistance begins, explained Blase, with the management teams becoming increasingly transparent about the company’s direction and the skills it will need. Each year starts with Stephenson giving a town hall speech for all the top managers of AT&T. “The idea is to be totally transparent with our employees about where the business is heading and what the challenges will be,” explained Stephenson.
That message gets filtered down through managers, so that eventually every employee has a broad understanding of the company’s objectives for the next twelve to fourteen months, “and where this company is going for the next five to ten years,” added Blase. “We start in January, and by July everyone has the message.” Ideally, he said, most employees will say, “I get it, I want to be part of this. I am one of the three hundred thousand. How do I get to be part of this?” Others, though, will say, “You know what? I’ve had my thirty-five years; it’s time to go. I am not ready to learn anything new.” So about 10 percent of AT&T’s workforce drops out each year.
Blase added:
We do not have enough internal people with the skills to be able to lead effectively through this change or the technical knowledge of what we are selling or the technical knowledge of what is behind what we are selling. So we hire thirty thousand [employees] off the street each year. We fill another thirty thousand jobs through rotations and promotions. It costs about two thousand dollars just to hire someone, so our preference always is to use our internal employees. It is more cost-effective and will generate more employee engagement and productivity, which means employees will go the extra mile so customers will be served better and shareholder value will increase. The companies with the most highly engaged workforces earn three times those with less.
But this has meant placing a lot more lifelong learning demands on a lot more employees. Most employees “embrace what we’re trying to do,” said Blase. “They say, ‘Just give me the tools, point me in the right direction, help me make [the transition] seamless, and make it cost-effective and make it mobile and make it Web based, so I can do it on my own time, and make it flexible and make the training in a format that I can learn quickly and effectively.’”
Added Donovan: “We have employees here who want to make the pivot—employees who built this company that they would die for—and we need to give them the opportunity to pivot. Many of these were traditionally blue-collar folks, who just finished high school, and we need to constantly retrain them to work in a networked house.”
To act on that bargain, AT&T five years ago asked all 107,000 of its managers (which is how it categorizes all its professional employees not covered by union contracts, regardless of whether they are a supervisor or an individual contributor) to set up internal profiles similar to LinkedIn accounts, detailing their job experiences, skills, education, certifications, and specialties. Today, 90 percent of the 110,000 managers have done so. Now when a new job opens up, the first thing Blase’s team does is check those profiles for internal candidates who have the necessary skill sets. At the same time, the company posts the hot new jobs and identifies where they are located, the exact skills needed to get those jobs, and how to get the training for those skills.
To help deliver on the latter, AT&T partnered with many universities—from Georgia Tech to Notre Dame to Oklahoma to Stanford to online universities such as Udacity and Coursera—to provide affordable graduate and undergraduate degrees or just specialized training for each one of the skills it needs. AT&T’s only requirement is that you take the courses on your own time, but the company will reimburse your tuition up to eight thousand dollars a year (or more, for certain courses) and thirty thousand dollars over your lifetime at the company.
Then, to make sure that money went as far as possible, AT&T pressed universities to create a menu of online learning opportunities that would fit its budget. This approach has driven a lot o
f innovation in education, most notably the partnership between Udacity, AT&T, and Georgia Tech to create an online master’s degree in computer science for $6,600 for the entire course—as compared with the $45,000 it would cost for two years on campus at Georgia Tech. Coursera has partnered with Johns Hopkins and Rice to create a similar certificate in data science.
This is driving down the cost of education for everyone. The education “pie just got bigger,” said Blase. “We can now assist you to get the job of your dreams.”
That’s intelligent assistance. “We do $250 million of training a year,” said Blase.
A lot is teaching people to climb poles, install services, and run retail stores, but now a lot more is in data science, software-defined networks, Web development, introduction to programming, machine learning, and the Internet of Things. And if you want to take a general STEM [science, technology, engineering, math] course that is not part of our program, we will pay for it, too. If you want to learn, we’re all in, because, again, it leads to more engaged employees; that equals better customer service, more loyal customers, and more shareholder value. We did not have anything like this when I was growing up in the company.
These supports are for jobs paying sixty thousand to ninety thousand dollars a year.
The company registers all certificates and degrees earned by employees in their company profiles and can easily search them through big data tools. And if you show that you are motivated to get these advanced degrees and certificates, said Blase, “we will give you first crack when a job opens. People need to know that if I am clearly motivated to learn, I am going to get rewarded.”
The way the system works, said Blase, is:
Say I am a manager and I have ten openings for technical jobs. I come to HR and they say you have to look [internally] first. Then you look through the online profiles and you find people who have shown or demonstrated a desire to get into one of these capacities. So HR will pull a bunch of names capable for those ten openings, [including] someone who has a majority of the skills. You pull the aspiring people and the perfect fits, and we ask the person who is doing the hiring to give these people who are trying a shot.
Because, said Blase, those employees will then tell their story to other employees: I played by the new rules and got rewarded. As he put it:
It is a contract between the company and employees. It’s a new bargain. If you want to get an A in your performance review, now you have to do the “What” and the “How.” The “How” is that you get along with people, you achieve results by effectively partnering and teaming and leading change through [and with] others and don’t just sit in your cubicle. The “What” is that you are not only proficient in your job but that you are reskilling to improve your capacity, continuing to learn, and that you are aspiring to go beyond where you are. Maybe you’re a salesperson and you’re making yourself more valuable to the company by getting [to know] the technical side as well. You’re not just selling products but understanding how our network works. Our best employees have it down and they know it is the What and the How.
The new social contract, Donovan added, is that
you can be a lifelong employee if you are ready to be a lifelong learner. We will give you the platform but you have to opt in … Everyone has a personal learning portal and they can see where the endpoint is [for whatever skill set they are aiming to acquire] and the courses that will get them there. You can pick a different future and how to get there. You can be anything you want to be in this system. But again, you have to opt in. The executive’s role here is to define the vision for the future. The company’s responsibility is to provide the tools and platforms for employees to get there, and the individual’s role is to provide the selection and motivation. We need to make sure that anyone who leaves here [does not do so] because we did not provide them the platform—that it was their lack of motivation that did not make it happen.
AT&T is a big whale. When it moves down this pathway of education to employment, it creates a huge wake. As Blase put it: “Now, universities are modifying their behaviors to meet us in the marketplace. We are writing a new blueprint.” If universities are paying attention, they will start creating more degrees and certificates “that are profitable for them but are cost-effective for this model.”
Donovan is certain that this new social contract is raising both the company’s average skill level and its morale. “What we have done is to take our best and make it our average,” he said, “and our average is now right up there. Our cycle times for [new ideas] are much faster now. Anyone who finds a solution, we can scale it through the company. Our employee engagement surveys showed a 30 percent improvement in lost sick days in one year. People are calling in sick less because they are feeling more empowered, more of a sense of ownership, and more connected.”
Jump-Starting the Curriculum
The AT&T model has wide-ranging implications for the entire world of education. Consider Udacity, which built the online low-cost master’s degree in computer science with Georgia Tech. Today, the business it created with AT&T enables it to offer the same intelligent assistance to the world at large and plant the seeds of the real revolution in education.
Udacity was founded by German-born Sebastian Thrun, formerly an artificial intelligence professor from Stanford and an expert on robotics. Thrun likes to recall his first meeting with Randall Stephenson at AT&T headquarters in Dallas: the two of them sat on the floor together in Stephenson’s executive suite so Thrun could use his laptop to sell the AT&T exec on how mini online courses, or nanodegrees, that could teach the latest technology skills could elevate the AT&T workforce. Stephenson got up off the floor after the demo and signed him up on the spot. One of the things that Thrun learned from teaming up with Georgia Tech to create the $6,600 online master’s in computer science was that the course did not cannibalize Georgia Tech’s far more expensive campus-based master’s. It turns out there were two different markets—one for people who want a campus experience and the other for people starved for lifelong learning that they could do in their spare time at a price they could afford. “The average age of the students for our online course is thirty-four and for the campus course is twenty-three,” he explained. The demand for more lifelong learning platforms was clearly out there. People get it. So today Udacity offers nanodegree programs on building websites, introduction to programming, machine learning, developing apps for Android mobile devices, and developing apps for Apple mobile devices, among others.
But here is where things get really interesting: Udacity now develops some of its courses with the help of working Google engineers. So, for instance, in October 2015, Google released the basic algorithms for a program called TensorFlow for public consumption by the open-source community. TensorFlow is a set of algorithms that enable fast computers to do “deep learning” with big data sets to perform tasks better than a human brain.
“By January 2016 we had a course online on how to use the TensorFlow open-source platform to write deep learning algorithms to teach a machine to do anything—copyediting, flying a plane, or legal discovery from documents,” explained Thrun. This is a huge new field of computer science. TensorFlow was released into the wild in October, and by January, Udacity, working directly with Google engineers, was teaching the skill on its platform. “We can now update your skills at the pace of Moore’s law, at the pace of the industry,” explained Thrun. The traditional “academic world cannot do that.” It would probably take a year for a university to have such a course up on TensorFlow, and for many it would take a lot longer.
Udacity has developed a stable of on-demand freelancers around the world whom it employs to grade the work of its online students—and the students also grade their graders. “I can hire a thousand graders in a week from around the world,” said Thrun, “try them out, find the two hundred best, and let the other eight hundred go.” It is a fast way to get high quality. There are Udacity freelance graders who make several thousand d
ollars a month grading computing projects—like how to build a map from Google’s GPS—sent in from students around the world. “We had one project grader who made twenty-eight thousand dollars one month,” said Thrun. “The gig economy is moving up. It’s not just about TaskRabbit errands anymore.”
And Udacity is not just providing intelligent assistance for companies such as AT&T. Its platform is creating intelligent assistance for “the start-up of you”—whoever and wherever you are. In the fall of 2015, I found myself in a small conference room at Udacity’s Palo Alto headquarters interviewing—via Skype—Ghada Sleiman, a thirty-year-old Lebanese woman, who was taking Udacity’s online course to advance her skills in Web-page design. She explained that she was sitting at her home in Beirut taking a course from a company in Palo Alto in order to better serve her clients, most of whom she has never met in person, in Australia and the United Kingdom.
“I studied graphic design at the American University of Science and Technology in Ashrafiyya,” a neighborhood of East Beirut, she explained to me. “After college I was looking for a course on Web design, and I found Udacity and decided to try it out. I started last year [in 2014]. I used to learn [on the Web] only through tutorials.” But she found that with Udacity’s platform “there was a sense of community, and I could communicate with other people, so it was more interesting and more interactive.”
Thank You for Being Late Page 25