War of Intelligent Machines

Chapter 18: Grasping the Key Point

Chapter 18: Focusing on the Key Point

After Jiang Yue Sheng returned home, he felt as though a heavy stone was pressing down on his chest. He couldn't bring himself to eat or drink.

He had been thinking nonstop about a question: why could people easily identify the camouflaged robots at the test site with just a glance, while the robots themselves could not?

In fact, at the test site, no matter how the robots altered their camouflage—whether individually or in groups—we humans could spot them effortlessly. It was like seeing a horse grazing, then being able to identify a galloping horse or a rolling horse.

This was because our brains detected many features that robots couldn’t.

In other words, they needed to focus on recognizing the robots’ specific characteristics. If they could identify enough features of a robot, no matter how it disguised itself, it would be instantly recognizable.

With that realization, Jiang Yue Sheng felt that he needed to completely rethink his approach. Instead of focusing on environmental modeling, which was ever-changing and could never keep up with real-time variations, the key was to model the robot’s features. The environment could change, but the robots remained relatively consistent—if they could identify enough of a robot’s features, it would be “unbeatable,” no matter how the environment changed.

He suddenly recalled the AI model, and quickly opened it on his phone, typing: How do surveillance systems distinguish humans from other moving objects?

The answer came quickly:

Identifying contour features, including the ten positions of the body.

Identifying structural features, including the relative positioning and proportions of body parts.

Identifying motion features, including standing, walking, and running.

Identifying facial features, including the distance between pupils, the width between ears, and the distance between the tip of the nose and the brow.

Identifying movement contours, such as leg swing angles, the angle between legs and torso, arm angles, and the angle between arms and torso.

Identifying motion structure features, such as the relationship between the head, upper arms, and legs.

Seeing this, Jiang Yue Sheng immediately had an idea. Since robots were designed to mimic humans, most of the features used to recognize people could be transferred directly to robots. All he had to do was help the robots distinguish between humans and robots.

With that in mind, he entered the programming section of the AI model and typed: Please write a program to distinguish humans from other moving objects.

Seconds later, the AI model provided a complete program with over 5,000 lines of code.

The next day, Jiang Yue Sheng gathered the members of the team and presented his new idea.

“You’re completely overturning the previous plan, aren’t you?” Shi Jun asked, confused.

“Yes. Although the previous plan was mine, I now think it wasn’t a good one. It was a temporary idea that couldn’t solve the core issue. The solution I’m proposing today is the logical, first-principle approach.”

“Does that mean all the work we did before was for nothing?” Shi Jun said helplessly.

“Unfortunately, most of it was. But if we continue down the same path, the future workload will be much higher than if we adopt the new plan. So, it’s better to stop now before we waste more time.”

“I’m familiar with programming for human recognition, and I support your new idea,” one of the team members said, raising their hand.

“Well, I’m not that familiar with this, so I’ll assist you guys,” Shi Jun said, sounding somewhat discouraged.

The group leader was surprised when he heard that Jiang Yue Sheng had completely discarded the previous approach. However, after listening to Jiang Yue Sheng’s explanation at the team meeting, he agreed that Jiang’s reasoning made sense. As an experienced programmer, he understood the importance of program architecture. The architecture was the foundation, the roadmap of the program’s future. If they followed the wrong path, it would only lead them into darkness.

After full discussion, the group leader made the final decision: they would follow Jiang Yue Sheng’s new approach, and he would personally act as the architect.

Two months later, the program based on the new architecture was completed. It ran smoothly on the simulation system, so they began arranging the scene tests.

This time, the scene team made the challenge even more difficult by enhancing the realism of the camouflage and adding some tests involving partial obstructions, hoping to see if Jiang Yue Sheng’s new approach could identify robots’ partial features.

The test began. The robots started approaching the building complex, and the enemy robots, positioned ahead, had applied partial camouflage. However, they were quickly detected and eliminated.

This confirmed that partial feature recognition was effective.

The robots continued to advance deeper into the complex, where the enemy robots had grouped together in complex camouflage formations, attempting to blend in with the surrounding architecture.

This was the same stage where the previous test had failed, and everyone holding their breath wondered whether this time would be any different.

As the test robots advanced, they slowed down, constantly rotating their heads, as if deep in thought. Then they raised their guns and started shooting at what they believed to be their targets.

With flashes of white light, the camouflaged robots, now properly identified, were all hit and fell.

“Success!” Shi Jun was the first to shout.

“We did it!” the group leader exclaimed, running into the testing area, visibly excited.

...

“Is it Jiang Yue Sheng?”

While Jiang Yue Sheng was working at his desk, he suddenly heard someone call his name. He quickly turned around and saw the director of the HR department smiling at him.

“Come with me,” the director said before turning and leading Jiang Yue Sheng to a nearby conference room.

“There’s a new position,” the director said as Jiang Yue Sheng sat down, getting straight to the point.

“Oh…”

“The R&D center at headquarters has an opening. The position is mainly responsible for developing software for the peripheral modules of police robots, with an annual salary of 600,000. Someone at the branch recommended you, so I’m here to ask if you’re interested. The job would be based at the headquarters in Qianhai.”

Jiang Yue Sheng paused, then reflected on what he had just heard. The salary was tempting, but this new position meant a lot of changes. He couldn’t help but think about the concept of focusing on the key points.

There was a video from a street interview once where a reporter asked a girl: “What type of boy do you like?”

“I like boys with beautiful hands.”

Many people might find this odd—does having pretty hands really matter?

This was actually about the idea of focusing on the key point.

In life and work, we often face problems. To solve them, we must first identify the key issue. As Chairman Mao once said, we need to focus on the main contradiction.

Clearly, whether someone has beautiful hands is not a key point.

Perhaps the girl knew this herself, and intentionally downplayed the importance of looks. A street interview is casual, and such answers can be quite charming. But when it comes to critical issues, it’s crucial to focus on what really matters—that’s why we were taught to identify the central idea in texts.

If you read an article but miss the main point, you’ve missed the core.

Some might say, “I don’t want to live so stressed out, constantly racking my brain. I just want to live a carefree life.”

Everyone has the right to choose how they live. But what’s the result of living carefree? It’s watching others succeed and feeling envious when they make money or get promoted.

There’s no such thing as perfection in the world. The hard work of today is for the glory of tomorrow, and the ease of today will lead to regret in the future.