The learning machine
The learning machine
The online commercial empire rests on a low-key approach to artificial intelligence
This Internet company Business empire has chosen a low-key path in the development of artificial intelligence
Amazon's six-page memos are famous. Executives must write one every year, laying out their business plan. Less well known is that these missives must always answer one question in particular: how are you planning to use machine learning? Responses like “not much” are, according to Amazon managers, discouraged.
Amazon’s six-page memo is so famous that executives every year You are required to write one page detailing your future business plan. But what’s less well-known is that each letter must answer a specific question: How do you plan to use machine learning? If your answer is "nothing to say," that answer is not allowed, according to Amazon management.
Machine learning is a form of artificial intelligence (ai) which mines data for patterns that can be used to make predictions. It took root at Amazon in 1999 when Jeff Wilke joined the firm. Mr Wilke, who today is second-in-command to Jeff Bezos, set up a team of scientists to study Amazon's internal processes in order to improve their efficiency. He wove his boffins into business units, turning a cycle of self-assessment and improvement into the default pattern. Soon the cycle involved machine-learning algorithms; the first one recommended books that customers might like. As Mr Bezos's ambitions grew, so did the importance of automated insights.
Machine learning is a way to achieve artificial intelligence , which mainly includes specific types of data mining, with the main purpose of predicting future trends. The idea began to take shape in 1999 when Jeff Wilke joined the company. Mr. Wilke is the second-ranking person at Amazon. He has formed a group of artificial intelligence experts mainly responsible for research on Amazon's internal workflow, with the purpose of improving employee work efficiency.
He placed scientists in various corporate departments and fixed the continuous cycle of self-evaluation and improvement as a default mode. This cycle was soon joined by algorithms; the first generation of algorithms could recommend books to customers that they liked. As Mr. Bezos’s ambitions grow, this fully automated algorithmic recommendation model becomes increasingly important.
Yet whereas its fellow tech titans flaunt
What do other tech giants have to show off
Their ai prowess at every opportunity—Facebook's facial-recognition software, Apple's Siri digital assistant or Alphabet's self-driving cars and master go player—Amazon has adopted a lower-key approach to machine learning. Yes, its Alexa competes with Siri and the company offers predictive services in its cloud. But the algorithms most critical to the The company's success are those it uses to constantly streamline its own operations. The feedback loop looks the same as in its consumer-facing ai: build a service, attract customers, gather data, and let computers learn from these data, all at a scale that human labor could not emulate.
Technology giants seize every opportunity to show their strength in AI: Facebook has launched facial recognition software, Apple has its voice assistant Siri, and Google has launched self-driving and Alpha Go. Compared with these companies, Amazon has chosen a low-key path in machine learning. Alexa (Alex) is an artificial intelligence service launched by Amazon. Its main competitor is Apple's Siri. Relying on Alexa's cloud platform, Amazon can provide users with predictive services. The algorithm behind this artificial intelligence is quite unique. It can continuously streamline its operation process, but the feedback loop of this AI service is similar to its client AI: launch a service, attract target customers, collect user information, and let Computers learn from this data and process it on a scale that is beyond human reach.
Mr Porter's algorithms
Mr Porter's algorithms
Consider Amazon's fulfilment centers. These vast warehouses, more than 100 in North America and 60-odd around the world, are the beating heart of its $207bn online-shopping business. They store and dispatch the goods Amazon sells. Inside one on the outskirts of Seattle, package shuttle along conveyor belts at the speed of a moped. The noise is deafening—and the facility seemingly bereft of humans. Instead, inside a fenced-off area the size of a football field sits thousands of yellow, cuboid shelving units, each six feet (1.8 meters) tall. Amazon calls them pods. Hundreds of robot shuffle these in and out of neat rows, sliding beneath them and dragging them around. Toothpaste, books and socks are stacked in a manner that appears random to a human observer. Through the lens of the algorithms guiding the process, though, it all makes supreme sense.
We can take a look at Amazon's "fulfillment centers." They are actually large warehouses, with more than 100 in North America and more than 60 scattered around the world. It can be said that these warehouses are the powerful heart of the company, and they drive Amazon’s online shopping trade worth $207 billion. These warehouses are used to store and distribute goods before Amazon sells them to customers. In a warehouse located on the outskirts of Seattle, the conveyor belt transports packaging supplies at the speed of a locomotive. It is difficult to hear a little noise, and these facilities are basically fully automated. In a fenced-in area, an area about the size of a football field contains some yellow square-shaped shelves. Each shelf is about 1.8 meters high. Amazon calls them "small warehouses." These "warehouses" are neatly arranged in a row, and hundreds of robots shuttle between them, moving them out and in. From a human perspective, these items, such as toothpaste, books and socks, are placed randomly on the shelves, which is really incomprehensible. But under the guidance of the algorithm, this process appears extremely reasonable.
Human workers, or “associates” in company vernacular, man stations at gaps in the fence that surrounds this “robot field”. Some pick items out of pods brought to them by a robot; others pack items into empty pods, to be whirred away and stored. Whenever they pick or place an item, they scan the product and the relevant shelf with a bar-code reader, so that the software can keep track.
Human employees , or "human partners" as Amazon calls them, mainly provide auxiliary services to robots. Their workplace is located at the platform between the fences, and the inside of the fence is the so-called "robot zone". The robots are constantly moving small warehouses. Some employees take the goods from them, and some put the goods back into the empty warehouses. But whether employees take it out or put it back, they will use a barcode scanner to scan the product and the corresponding shelf, so that the software system can record the running path of the product.
The man in charge of developing these algorithms is Brad Porter, Amazon's chief roboticist. His team is Mr Wilke's optimization squad for fulfilment centers. Mr Porter pays attention to “pod gaps”, or the amount of time that the human workers have to wait before a robot drags a pod to their station. Fewer and shorter gaps mean less down time for the human worker, faster flow of goods through the warehouse, and ultimately speedier Amazon delivery to your doorstep. Mr Porter's team is constantly experimenting with new optimizations, but rolls them out with caution. Traffic jams in the robot field can be hellish.
Brad Porter is the primary developer and manager behind these algorithms He is also the chief robotics scientist at Amazon. The team he assembled was an optimized version of Mr. Wilke's team, primarily serving the fulfillment center. Mr Porter's main focus is on how to close the gaps between small warehouses and reduce the time human workers spend waiting at their stations for robots to deliver goods. For human workers, fewer and smaller gaps mean shorter loading and unloading times, faster cargo transportation processes, and faster delivery services. Mr. Porter's team has been experimenting with new optimization strategies, but every rollout has been very cautious, because traffic jams in the "Robot Zone" are a very serious and terrible problem.
Amazon Web Services (aws) is the other piece of core infrastructure. It underpins Amazon's $26bn cloud-computing business, which allows companies to host web-sites and apps without servers of their own.
Amazon Web Services (AWS) is another component of its core infrastructure. Its existence sustains Amazon's $260 billion cloud computing business. Using this network system, companies can open their own websites or develop their own applications without a server.
aws's chief use of machine learning is to forecast demand for computation. Insufficient computing power as internet users flock to a customer's service can engender error and lost sales as users encounter error pages. "We can't say we 're out of stock," says Andy Jassy, ??aws's boss. To ensure they never have to, Mr Jassy's team crunches customer data. Amazon cannot see what is hosted on its servers, but it can monitor how much traffic each of its customers gets , how long the connections last and how solid they are. As in its fulfilment centres, these metadata feed machine-learning models which predict when and where aws is going to see demand.
AWS's machine learning The primary use is forecasting computing needs. When Internet users flood into the client, the lack of computing power will cause many errors, such as users entering error pages and transactions having to be cancelled. "We can't say we don't have inventory." Andy Jassy, ??the boss of AWS, said that in order to ensure that this network system never goes wrong, his team collects and analyzes a large amount of customer data. While Amazon has no way of knowing what's on its servers, it can detect how much traffic customers receive, how long their connection to the server lasts, and the quality of that connection. In Amazon's execution centers, machine learning models rely on the input of these metadata to run. The function of these models is mainly to predict when and where AWS systems are likely to generate computing needs.
One of aws's biggest customers is Amazon itself. And one of the main things other Amazon businesses want is predictions. Demand is so high that aws has designed a new chip, called Inferentia, to handle these tasks. Mr Jassy says that Inferentia will save Amazon money on all the machine-learning tasks it needs to run in order to keep the lights on, as well as attracting customers to its cloud services. “We believe it can be at least an order-of-magnitude improvement in cost and efficiency," he says. The algorithms which recognize voices and understand human language in Alexa will be one big beneficiary.
One of AWS's largest customers is Amazon Own. At the same time, the demand for AWS from other Amazon businesses is also focused on its predictive capabilities. Due to the huge amount of calculations, researchers designed a new chip for AWS to handle these tasks. It is called Inferentia. Mr. Jesse said the chip will save Amazon a lot of money on various machine learning tasks while attracting more customers to choose its cloud services. Mr. Jesse also said that "Inferentia will bring an order of magnitude improvement to the company's cost efficiency." Alexa, which can recognize sounds and understand human language, will bring endless benefits to its own algorithm development.
The firm's latest algorithmic venture is Amazon Go, a cashierless grocery. A bank of hundreds of cameras watches shoppers from above, converting visual data into a 3d profile which is used to track hands and arms as they handle a product. The system sees which items shoppers pick up and bills them to their Amazon account when they leave the store. Dilip Kumar, Amazon Go's boss, stresses that the system is tracking the movements of shoppers' bodies. It is not using facial recognition to identify them and to link them with their Amazon account, he says. Instead, this is done by swiping a bar code at the door. The system ascribes the subsequent actions of that 3d profile to the swiped Amazon account. It is an ode to machine learning, crunching data from hundreds of cameras to determine what a shopper takes. Try as he might, your correspondent could not fool the system and pilfer an item.
In terms of algorithm exploration, the company’s latest achievement is Amazon Go is a grocery store without cashiers. Hundreds of cameras in the store monitor customer behavior from above at all times and convert the collected visual data into three-dimensional user information. The purpose of this data is to track the arm movements of customers when picking up goods. In this way, the algorithm system can know which products the customer has taken, and automatically send the bill for these products to the customer's Amazon account when the customer leaves the store. Dilip Kumar, the boss in charge of Amazon's Go project, emphasized that the purpose of the system is to track customers' body movements and does not use facial recognition to identify customers' information to connect to their Amazon accounts. The system is an ode to machine learning, gathering information from hundreds of cameras to determine exactly what customers took. Maybe you're trying to steal an item, but these camera systems aren't easily fooled.
Fit for purpose
Tailor-made
ai body-tracking is also popping up inside fulfillment centers. The firm has a pilot project, internally called the “Nike Intent Detection” system, which does for fulfilment- center associates what Amazon Go does for shoppers: it tracks what they pick and place on shelves. The idea is to get rid of the hand-held bar-code reader. Such manual scanning takes time and is a bother for workers. Ideally they could place items on any shelf they like, while the system watches and keeps track. As ever, the goal is efficiency, maximizing the rate at which products flow. “It feels very natural to the associates, ” says Mr Porter.
Artificial intelligence motion tracking also has a role within the execution center. Amazon has launched a pilot program, known internally as Nike Intent Detection, that operates in fulfillment centers on the same principle as Amazon Go: tracking items as they are taken out and back on shelves. The idea is mainly to eliminate the previous hand-held barcode scanners, because such input work wastes employees' time and is very cumbersome to operate. Ideally, employees could place items on any shelf, with the system monitoring and tracking them. Amazon’s goal has always been to increase efficiency and maximize the flow of products, a process that, in Mr. Porter’s words, “feels natural to all of our human employees.
”
Amazon's careful approach to data collection has insulated it from some of the scrutiny that Facebook and Google have recently faced from governments. Amazon collects and processes customer data for the sole purpose of improving the experience of its customers. It does not operate in the gray area between satisfying users and customers. The two are often distinct: people get social media or search free of charge because advertisers pay Facebook and Google for access to users. For Amazon, they are mostly one and the same (though it is toying with ad sales). Where regulators do raise concerns is over Amazon's dominance in its core business of online shopping and cloud computing. This power has been built on machine learning. It shows no signs of waning.
In terms of data collection, Amazon has chosen a very cautious path. Therefore, compared with Facebook and Google, relevant government departments have much less scrutiny on Amazon, and some parts can even be exempted. The main reason is that Amazon. The user information collected and processed is only used to improve the user's operating experience, and there is no gray area between meeting the needs of users and consumers. The difference between data users and producers (consumers) is usually clear: People are able to use social media or free search engines because advertisers pay Google and Amazon to have their ads reach consumers (although to Amazon, they are essentially the same person). Amazon doesn’t really care about advertising revenue). But Amazon also faces some regulatory concerns, such as its monopoly in two major business areas: online shopping and cloud computing. But this status is built on the strength of machine learning, and there is no sign that they are in decline.