The Forbes website posted an article by Peter High, president of Metis Strategy, a business and IT consultancy, interviewing Facebook Chief Technology Officer Mike Schroepfer. In the interview, Skropov explained the three major innovation pillars of the company's previously announced 10-year development plan: network connectivity, artificial intelligence, and virtual reality.
The following is the main content of the article:
At the Facebook F8 conference earlier this year, the company unveiled three major innovation pillars that will shape its vision for the next decade: Internet connectivity, artificial intelligence, and virtual reality. Its Chief Technology Officer, Mr. Skropf, is responsible for leading these three projects. Although the vision is ten years, the company has actually made significant progress in three areas.
Facebook’s progress in artificial intelligence can be seen everywhere, from the company’s news feed to the way people mark things. Innovations in virtual reality are best shown by the Oculus Rift. Recently, the company has taken an important step in the network connection: Acquila set sail in Arizona, and its flight altitude will exceed commercial aircraft and meteorological layers. Facebook’s goal is to bring basic networking services to developing countries through the V-plane, which spans the Boeing 737 but weighs less than 1,000 pounds.
The Forbes reporter met with Skoropf at Facebook’s headquarters in Menlo Park recently and discussed the three major innovation pillars of the social network and many other topics, including the company’s recruitment methods and how it is maintained. Innovative advantages, and design logic behind its headquarters, one of the largest open space office buildings in the world.
The following is an interview summary:
Q: At the F8 Developer Conference held earlier this year at Facebook, you put forward three pillars of innovation. Can you introduce them again?
Skrotpoff: We can say in the industry is a very special existence. We are very happy to share our vision of the decade and development roadmap. We have broken it down into three core areas:
· Internet connection, connecting nearly 4 billion people in the world with no Internet access (a large part of the global population);
Artificial intelligence to solve some of the core issues and build a truly intelligent computer system;
Virtual Reality and Augmented Reality create a promising next-generation computing system that allows people who are separated by miles to feel like they are in the same room.
Q: Facebook actually has no shortage of resources and talents. How did you pick out the three pillars instead of picking out other things?
Skopov: Many of this came directly from Facebook CEO Mark Zuckerberg, stemming from the mission of making the world more open and interconnected. I just think of this as using technology to connect people. We sat down together and discussed: "If that is our goal, it is the best thing we can do. Then what are the major problems in this world?" As you begin to analyze this, the answer naturally comes out.
The first problem is that if a large part of the world’s population does not even have basic networked services, then it is a fundamental issue. Then, through discussion and analysis, you will realize that the problem is technically a solution; some things can significantly reduce the cost of deploying network infrastructure, which is a major constraint. In fact, it is only an economic issue.
Once people are connected to the Internet, they will encounter problems that you and I have encountered. That is information overload. There is too much information nowadays, but my time is limited, so I may not be able to get the most valuable information. After that, we realized that the only way to effectively solve this problem is to start building intelligent systems based on artificial intelligence, so that artificial intelligence can be my real-time assistant 24 hours a day, to ensure that I will not miss any information that is important to me. I will not spend my time on unimportant matters. We know that the only way to achieve this on a large scale is through artificial intelligence.
Well, the problem of networking and getting value information can be solved. However, many of us have friends and relatives who are not always with us. We cannot always go with them to witness the important moments in their lives. The advanced technology we can use today is video calling. If I want to capture the lives of my children and remember them forever, video calls are the best tool available. The problem is, if I want to appear on the scene to record those times, I feel like I'm still on the scene when I'm reviving 20 years later? This is where virtual reality can come in handy. It allows you to experience it all after wearing a helmet, making you feel that you are really in a real place thousands of miles away, so that you can go wherever you want to go.
Q: How do you see the long-term goals that will take some time to come by, and the few short-term projects to be commercialized and ready for commercialization?
Skopov: For the 10-year development plan, I would like to say that the most important thing is to be patient when things go wrong, because the process of making any big event will not be easy. I also started a company in the past. The process was never smooth. From the outside world, my work on Facebook may have been going smoothly all the time, but in fact it will face all kinds of ups and downs along the way. It is important not to feel discouraged when the encounter is not right.
In addition, if you can reach milestones on the way forward that can help you know you are making progress, you should do it. The artificial intelligence function is very simple because the team has deployed a lot of things to significantly improve people's daily Facebook experience. They can be simple technologies like helping you better organize photos, letting you not lose important life photos, or they can be more basic technologies, like the assistive technologies we introduced earlier this year, it can be Visually impaired people who cannot see billions of photos uploaded to Facebook daily generate and read photo instructions.
At the same time, we are studying more difficult issues. For example, how do you teach computers to digest unstructured data such as Wikipedia articles and then answer data-related questions? This may sound simple, but it is at the forefront of artificial intelligence research: inferring and understanding undigested and unoptimized things without anyone providing you with a good, valuable format. From this perspective, the problem is very far-reaching.
In the field of virtual reality, we want to bring outstanding product content to the market to attract you to Buy Oculus Rift helmets at Best Buy. We know that in 2016 this is a relatively niche market. Compared to Facebook’s current user base of more than one billion, its audience is still very small, but we hope that content facing virtual reality will be outstanding year after year. We will roll out system upgrades every few years, and as these systems get better and cheaper, VR users are expected to reach hundreds of millions in a decade, not just millions. You must be patient with the process of accumulating your audience and not be disappointed by the small number of users in the short term.
We have achieved some milestones in network connectivity, such as the Acquila aircraft that sailed a few weeks ago. But that was only the first flight of our plane. It is far from ready to implement its ultimate goal, and its ultimate goal is to provide network connectivity to people in areas where network infrastructure is too expensive.
You have to try to have a long-term vision, be patient while things may be problematic, and at the same time take some intermediate steps on the way forward, make progress and increase value, and see how things progress. At least this is how I view long-term and short-term goals.
Q: Solving the problem you are aiming for not only requires brainpower, but also requires talents with a wide range of industry backgrounds. You need talents with STEM (Science, Technology, Engineering, and Mathematics) education backgrounds, talents with advanced neuroscience expertise, and so on. And, in Silicon Valley, the talent competition is so fierce. What do you think about talent acquisition?
Skoplov: To be honest, this is where the real fun of work lies. Today, we are indeed facing challenges in the search for talent, but when you find a group of interdisciplinary and multi-disciplinary talent, magical things happen. Let's talk about the Acquila aircraft. You have to have high-end composites, electrical systems and motors. An important part of this is the use of a new generation of battery technology based on solar cells. Next, we need a communication system that can transfer the network up and down the plane. This means that we need to do free-space optical communication, and we need laser communication equipment, laser transmitters and receivers. You need a software system to help control the flight of an unprecedented aircraft. We need to do some simulation on what it looks like. You need to machine learning software, hardware, electronics, aerospace, and materials science. You need to combine all of this.
When you combine the talents in these fields into a small team—grouping dozens of people—and let them clearly target this goal, many wonderful things will happen. Our company's core strategy, our technology, is about getting people together.
Q: How much freedom does the employee have in choosing the field of exploration?
Skoplov: They have a lot of freedom. From the moment I came to Facebook, the basic principle was to get smart people who were at the top level in their fields and let them freely choose where to work. If you let them engage in the things that make them most excited, they can do their best. If they wake up early in the morning and can't wait to solve the problem at hand, they will be able to bring much better work results.
Of course, sometimes we try to convince some employees to try to do something else, or explain to them what is the most important, but I basically can make the right people appear in the right place, after that I only We need to clearly articulate our ultimate goal: "This is what I want you to do. You can bring in a lot of contributions. Think about what is most exciting to you, and then concentrate on doing it." Then we can start to do something together and develop technologies together. Therefore, we have a lot of freedom in the selection of work content.
Q: What do you think about building a team outside of the company and building an ecosystem outside the company?
Skoplov : This is an area where we are trying to do a lot of innovation. Looking back at the creation of the Open Compute Project five or six years ago, it feels like everyone has accepted open source as a good way to develop software. I used to work at Mozilla, when we developed a browser through open source. Almost every company has Linux running - it's open source; we all make contributions to the same Linux kernel, but we don't do it for hardware development.
We discussed that, "Let's do that together to develop hardware." So our data center designed its own buildings, racks and servers - everything was open and people could collaborate. Today, the entire industry is collaborating, including Google. We also adopt the same strategy in the area of ​​network connectivity, such as the laying of telecommunications infrastructure. If we can get the entire industry involved, then ultimately everyone will benefit, because you can share a variety of core IP, you can develop on the same components, you can achieve economies of scale and gain productivity, products have become more It's cheap. It can also be widely used.
As for artificial intelligence research, we are actively making public and open source our research results. We have participated in all major conventions. I recently evaluated with the team their major progress. They developed a product called "memory network" a few years ago. This product is a way to attach short-term memory to the neural network of the brain. It is a function that we have not had before. That was the result of the 2014 research. Since then, there have been many research papers that have cited the work. Every month there will be new papers showing new enhancements to the technology that will improve some of the basic question and answer benchmarks.
Comparing the results created and accumulated by the entire industry, and if we go it alone, you will find that there is a huge difference because we can quickly advance research and development based on the work that has just emerged, two years faster than we alone do. time. Basic technology can benefit many people outside of Facebook. As long as we are within our capabilities, we will be happy to share that technology.
Q: How do you keep yourself up-to-date with the latest innovations happening inside and outside the company?
Skoplov: I think this is where I feel I might have the best job in the industry. First of all, I will read anything I can see. And more wonderful than that is to sit down with the team and talk about the work they are doing. This is the best part of my daily work so far.
Just a few weeks ago, I had a conversation with Facebook's research team and we had a one-day briefing on all the work they were doing. Yann LeCun, who wrote many papers on the neural networks in the 1990s, is leading the team to his vision for the company. The team is not only evaluating the work they are doing. Because they are open and open source to their research, they are also evaluating other people's works based on their research results to solve similar problems.
In addition, I will talk to some people who are developing the latest VR social applications to see what we can do in those areas. I also have the opportunity to talk with people in the industry or people who are interested in what I do at the Technology Conference. I have the opportunity to understand the current industry dynamics. It's interesting. To be honest, it is not easy to keep abreast of the latest innovations because there are a lot of interesting and attractive things happening all the time.
Q: After gaining the opportunity to circulate in your headquarters, I would like to ask: Have you ever thought of creating collaborative office space not only within these walls but also everywhere else on Facebook?
Skoplov: The building you live in is one of the largest single-storey office buildings in the world. There are 2,800 employees on this floor and no personal office is set up - you can see that everyone is on their desks. The purpose of this design is, as an experiment, to see how thousands of people can be put together in a room where we can push collaboration.
Obviously, one of the secret weapons we have is Facebook itself, and we have already provided it to others. Everyone is on Facebook all day, because we work here, it has become an excellent collaboration tool: You have a Facebook group, there is Facebook Messenger, there are a variety of good help you aggregate and understand all kinds of The way of information, for example, lets you know what your friends are doing recently, or what 16 different teams are doing at the same time. Those tools can help you understand these.
I think what many people tend to overlook is that the key to collaboration is that people can present their opinions and professional insights. They can have the opportunity to learn about another person and understand what the other person is doing. This can be cross-functional collaboration. For example, I'm doing machine learning. I want to understand what doctors are trying to do to discover patterns in drug discovery. The more I understand their problems, the more I can help them.
Our culture often brings about basic empathy, because on Facebook, you can not only see what colleagues at work are doing, you can also learn about them in daily life; children will go to school next week. Or they're back on vacation. It will bring about a sense of cohesion that I have not experienced in any other organization whose size is equal to ours.
Question: As the entire organization continues to grow and become a giant of technology, how worried are you with respect to maintaining a small company atmosphere and entrepreneurial spirit? Obviously, there are many innovative companies that were founded earlier than you were once brilliant, but after subversion they are in a dilemma.
Skopov: That is a question that we have been thinking about. There are good and bad places for the growth of the company. The advantage is that we can do exciting things and we have the best experts in all of these areas. If you are an engineer who is new to the company and wants to learn more about AI, aircraft or virtual reality, we can all meet your needs. This is good and it is a great benefit. However, it is a challenging task to allow a larger group of people to unite and work together to achieve a common mission.
I think this can be attributed to the few things we have been doing. First of all, you want the staff to have a real awareness of the ultimate goal. In large organizations, it may become federalized, that is, people do different things and their goals are not consistent. We are very clear about our mission - use technology to connect people - we clearly point out that if you are excited about this, it is good to join us; and if you do not feel excited about this, then you should choose an employer.
I like to think like this: A lot of our work is about cultural development. If you think of this kind of construction as a development system, then the way we work together is a system. We can spend time building such a system - the simplest thing is to provide training for new employees. For many companies, newcomer training is only a day and a half of paperwork. For us, it is a six-week high-intensity training camp. This design is designed to expose you to as many technologies as possible and to contact the company as much as possible.
By the end of the six-week training camp, no matter what kind of job you are engaged in, the number of company engineers you have met can reach hundreds, and your contacts are not just shallow contacts. During that time, you wrote some code and asked people to review it; you wandered around the company, talked to the engineers, and asked them about vulnerabilities.
So, when you finish the training camp, everyone will become very familiar with each other because no matter if you have been in business for 15 years or just graduated from college, all we are doing is knowing Facebook, which allows us to build relationships. . This relationship can spread to the entire company. You connect with those people in the group. And all of them have seen many people in the company. So when someone has questions about what the team is doing, he thinks. "Hey, I met Mary while studying this loophole. I would ask her, maybe she knows who to go to." This can build loose connections, but they are very important for building that kind of collective cohesion.
I can keep talking about this because we have created one project after another for building collective cohesion. In addition to training camps, we also implement job rotation: After entering the job for 19 months, you can take one month to complete rotation to another team. During this time you can learn new skills. You can work in a completely different work area. You can meet a group of new colleagues and continue the cohesiveness within the company.
Q: You've experienced it yourself. You exchanged positions with Sheryl Sandberg for a week.
Skoplov: Of course. I learned what it was like to be in her post and vice versa. We are trying to perform this job rotation in any place where it can be implemented. I think it can help. It is not easy to handle, so there are still many things to do, but we are focusing on this issue.
Q: In the recruitment process, how do you assess talent? Especially now companies are aiming for long-term goals.
Skrotpoff: The first thing you want to do is whether they have the skills needed to complete the job – we are now far more demanding on skills than ever before. In the past, we were mainly looking for talents with software engineering skills, and now we will also find talents in electrical engineering, mechanical engineering, or AI. For these positions, we have different skills assessment methods, but basically they are mainly based on certain technical verifications to see if they are top talent in their fields.
In addition, the ability to collaborate is also important to us. One part of the interview is the handling of collaborative issues. It consists of two parts: one is to see if they answer the questions correctly, and the other is to see if they can work together with another person to find the answers to the questions, because this is how the real world works. We want to ensure that this person has excellent communication skills and can work well with the team.