AI with powerful video understanding will become more common, and Ben Goertzel believes that AI will play an even more important role in the sport of competition. He predicts that within five years, the vast majority of the work of video coordinators will be taken over by AI. Once the AI learns the game and can apply it to the real game, the entire NBA will be turned upside down.

In other words, the managers of the team will be replaced by AI in the future. This day may come sooner than expected, after all, artificial intelligence is always developing at a faster depth than expected.

In recent years, data analysis, sports science, and a variety of new ways of measuring have made sport a huge scientific project. As a result, the former math nerds have become shining stars, and the athletic geniuses are just struggling to survive.

As statistics continue to grow, better technology is needed to help reduce noise.

"The NBA has built algorithms that integrate data from our SportVU system so that it can be used by every team in the league," said KenDeGe, NBA's senior vice president of IT applications

ARO said.

The SportVU system mentioned above consists of six cameras, three in each half, which take five images per second and then process them through a complex algorithm to extract the y, z positioning data of all objects on the field. Each picture is time-stamped and processed automatically by a computer.

The computer pushes the data to the live commentary stream and reports the game scene within 90 seconds. Basically, all the information will soon appear on the coaches and statisticians' computers or iPads.

In fact, the SportVU system was first created by Israeli scientists in 005 and is used in areas such as missile tracking and advanced optical identification. Then the system was used to track football matches in Israel, and then to the NBA.

The system can tell you not only Kevin Durant's shooting percentage, but also what his shooting percentage is when he dribbles once and twice, or what his shooting percentage is when a defender is walking away or five steps away from him.

When Rajon Rondo holds the ball for more than 5 seconds, what is the team's offensive efficiency? No one had been able to answer this question before, but with the help of SportVU, the answer was clear.

Every NBA court now has SportVU, a system that uses a fixed set of cameras to capture the movement of players and basketballs.

SportVU transforms what was once an unquantifiable and chaotic game of matches into stats that can be drilled deeper. That's where machine learning comes in.

Startups in Silicon Valley have long begun to apply artificial intelligence technology in the fields of healthcare, robotics, and enterprise software. Now it's finally the turn of the NBA teams.

SecondSpectrum, a Los Angeles-based company, uses artificial intelligence technology, including computer vision, to extract large amounts of data from NBA game videos.

With the help of this company, the Golden State Warriors were able to read game strategy and trend probabilities in a matter of seconds, which used to take months.

Kirk Lacob, assistant general manager of the Warriors, said there were a lot of videos on the team before, but there was no easy way to turn them into data. That is, until they meet SecondSpectrum.

The company, which has helped the Warriors, was founded by Rajiv Mahesaran and Yu-Han Chang, two AI professors at the University of Southern California. Chang also recruited at least 10 more engineers from MIT, where he graduated with a Ph.D. degree. SecondSpectrum has reportedly received early investment from Steve Ballmer and others.

Up to now, the company's customers have included more than a dozen NBA teams such as the Golden State Warriors and the Cleveland Cavaliers.

SecondSpectrum's software "learns" the precise movements of players, recognizing changes in their style of play and the operation of the basketball. In the case of pick-and-rolls, for example, the AI can identify whether the ball carrier needs or does not need a screen from a teammate, and whether the screener is ready to cut in after a pick-and-roll or cut out after a screen.

And how defenders should respond: block an opponent with the ball, follow the opponent, switch defenses, or make up for a pinch shot. The computer will also tell you what kind of shooting percentage each different option corresponds to.

For example, when the opponent is LeBron James, you must wonder how to defend against his pick-and-rolls, and whether it is more effective to squeeze past the cover and get close to the front. Or should you find a big man to sandwich James and how to deal with it in different situations.

All of this, the AI will tell you the answer.

It is said that an NBA team once privately revealed that with the help of artificial intelligence, the team found a strategy that could win the playoffs. Exactly how these new technologies affect the course of the game, every team will be tight-lipped about it.

But what is certain is that new technologies are having a greater impact on sports than ever before.

"Machine learning is becoming increasingly popular in data modeling," said Keith Goldner, a statistician at the NBA-NFL consulting firm.

Machine learning is an artificial intelligence technology that can improve itself after training on a large amount of data, without the need for a software engineer to edit a set of action instructions.

You can ask the AI to watch a series of matches, and then the algorithm will learn to recognize a certain tactic, and every time that tactic appears, the AI will automatically flag it. For example, the machine-driven tool mentioned above can classify and identify the different pick-and-roll tactics of NBA teams.

First learn to process data, then start helping the team develop tactics, and then start helping with on-the-spot command...... So the story at the beginning is going to appear again, starting with watching the video of the game, step by step, the AI is ready to replace the human coach.

The question is, how far is that day?

Goertzel admits there is still some way to go. "Video understanding is still not an easy task, sports are fast and complex, and it will take years to really solve this problem, but the tech giants are starting to focus on it," said Goertzel, who said that deep learning, which is already shining in the field of image recognition, will succeed in the field of video recognition.

Although the data is beautiful, what attracts fans is the performance of people on the field. And the manager has an unquantifiable, magical influence over the team. On the other hand, not everyone welcomes robots, especially in situations that seem to be human-specific.

But culture is always changing, and it's a trust-building process. There may be debate in the future about which is more important, AI or humans, but in most successful organizations, everyone already recognizes the value of both.

The scientists mentioned in the article do not want human coaches to disappear from the sidelines, but in the future, the cooperation between human coaches and machines will definitely become deeper and deeper. "In the medium term, AI will help managers deal with tactical matters. And at the end of the day, the manager may only have some social and strategic things left to do," Goertzel predicts.

As data analysis and game decision-making become more controlled by AI, managers will focus more on human-related tasks, and future generations of coaches may be more human than ever.

A great change is on the way, but we may miss out on the next Eric Spoelstra. (To be continued)