If Part 1, "Build Well," forged a keen "eye of observation" for us, and Part 2, "Manage Well," installed a rational "cerebral cortex" for this eye, then Part 3, "Use Well," is about truly opening this eye and deploying the core capability we call "the art of convergence" -- to observe, understand, and reshape the complex world we inhabit.
"The art of convergence" is our philosophical encapsulation of the intelligent computing center's core value. Its essence is to use massive computing power to converge a seemingly infinite, chaotic, and uncertain "space of possibilities" into one or a few optimal, certain "decision spaces" through computation and simulation.
Whether it is optimizing a production process, designing a new drug, or governing a mega-city, the fundamental challenge is the same. We face countless choices, countless variables, countless possible futures. And the mission of the intelligent computing center is to act as a powerful "convergence engine," helping us find the clearest and most reliable course through this fog.
This part will focus on the core application scenarios where the intelligent computing center empowers various industries. We will see how this engine sparks a profound revolution in efficiency and cognition in key areas such as urban governance, industrial manufacturing, and life sciences.
And the starting point for all of this, the most grand and complex application stage for the "art of convergence," is the city beneath our feet.
The city is humanity's greatest creation, and also the most complex governance challenge humanity faces.
A modern mega-city is a living organism. It has its own skeleton (roads and bridges), circulatory system (transportation and logistics), respiratory system (energy supply), nervous system (communication networks), and metabolic system (economic activity and waste disposal). Millions to tens of millions of people move and interact within this organism, generating astronomical amounts of multimodal, heterogeneous data streams every day.
In the past, governing a city was more like an experienced traditional Chinese doctor, using "observation, auscultation, interrogation, and palpation" -- observing road congestion, listening to citizen complaints, reviewing statistical reports -- to perceive the city's "pulse," and then relying on experience and intuition to prescribe "remedies" (policies). This model worked reasonably well when cities were small and simple. But today, faced with a complex mega-system with tens of millions of "cells," the limitations of traditional governance models are increasingly apparent. We often fall into a reactive pattern of "treating the head when the head hurts, treating the foot when the foot hurts," with policy-making often lagging behind the emergence of problems.
We can "see" problems -- see congested traffic flow on cameras, see excessive air quality readings on sensors, see overflowing complaints on citizen service hotlines. But this kind of "seeing" is passive, lagging, and fragmented.
The completion of the national intelligent computing center brings a fundamental paradigm shift to urban governance. What it aims to achieve is a leap from this passive "seeing" to an active, real-time, holistic "understanding," and ultimately to a forward-looking, simulative "foreseeing."
This is the true mission of the "City Brain."
5.1 The City Brain: The "Art of Convergence" in Everyday Governance
The concept of the "City Brain" is not new, but its implementation in many places still remains at the stage of a "city big screen" -- aggregating data from various departments onto a giant screen for visualization. This is merely completing the first step from "not seeing" to "seeing"; in essence, it is still a data dashboard, far from a true "brain."
The core of a true City Brain lies not in the "screen," but in the "computation." It must rely on the surging computing power of the intelligent computing center to converge the city's massive, disorderly perception data into precise, executable optimization decisions.
5.1.1 The First Level of Convergence: From "Data Flood" to "Digital Twin"
The starting point of the City Brain's work is processing the flood of raw data generated by millions of "nerve endings" (cameras, sensors, mobile devices) across the city. This itself is a formidable computing challenge.
Video Structuring: A city may have hundreds of thousands or even millions of cameras, generating petabytes of video data daily. If this data is merely stored for post-hoc forensics, it is "dead" data. The City Brain's first task is to perform real-time, large-scale structured processing of these video streams using the intelligent computing center's vision large models.
- Convergence process: Transforming unstructured pixels into structured semantic information. For example, a video from a traffic camera is no longer just a bunch of pixels, but is parsed in real time (illustrative format) as: "At intersection XX, the current westbound traffic flow is so many vehicles per minute, including some trucks and buses; the average speed is around such-and-such km/h; there are a number of pedestrians waiting to cross on the crosswalk..."
- Value: Only after completing this step does video data transform from "surveillance footage" into "intelligence" that can be analyzed and computed.
Multi-Source Data Fusion: Urban data is multimodal and heterogeneous. Besides video, there is geomagnetic coil traffic data, transit card swiping data from buses and subways, GPS trajectory data from ride-hailing services, temperature and humidity data from weather stations, public-opinion data from social media... These data are scattered across different departments and systems, speaking different "dialects."
- Convergence process: The City Brain needs to align, clean, and fuse this heterogeneous data within a unified spatiotemporal framework, constructing a "digital twin" of the city. In this virtual digital city, every vehicle, every building, and every pipeline has its precise digital avatar, with its state synchronized in real time with the physical world.
- Value: The "digital twin" is the foundation for all analysis, simulation, and optimization. It gives us, for the first time, a "God's-eye view" with which to "overlook" and "dissect" the entire city.
This first level of convergence is a stage for the intelligent computing center to display its "brute-force aesthetics." It requires massive GPU computing power to "digest" and "understand" this vast ocean of data. This is itself a highly computation-intensive "daily heavy task" and a typical scenario where the time-sharing leasing strategy can fully utilize nighttime valley computing power for batch processing.
5.1.2 The Second Level of Convergence: From "Possibility" to "Optimal Solution"
Once the "digital twin" is constructed, the true wisdom of the City Brain begins to emerge. What it needs to do is to find the optimal solution for the countless "multiple-choice questions" in the city's daily operation on this digital sandbox. This is the core application of the "art of convergence."
Let us take the most frequently mentioned example of "intelligent traffic signal timing" to dissect this convergence process:
Define the Problem (Determine the Convergence Target):
Traditional traffic lights are "static," switching according to a fixed cycle. The City Brain's goal is to make the city's traffic lights "dynamic," able to adjust their timing in real time based on actual traffic flow. The ultimate "convergence target" could be defined as: while ensuring pedestrian safety, minimize the average vehicle waiting time across the city, or maximize the traffic efficiency of the main roads.
Build the Model (Delimit the Possibility Space):
The brain first constructs a complex traffic flow model. This model considers the intersection's geometry, the number of lanes, traffic flow in different directions, vehicle types (buses, private cars), the influence of upstream and downstream intersections, and so on. For a complex intersection, the possibilities of green light durations and phase combinations are astronomical. If the linkage possibilities of thousands of intersections across the city are taken into account, this "possibility space" is one that the human brain can never exhaust.
Computation-Driven Simulation (Explore and Converge):
At this point, the computing power of the intelligent computing center comes into play. It uses advanced AI algorithms such as reinforcement learning to rapidly simulate and explore within this vast "possibility space."
- Simulation: An AI agent, like a traffic police officer possessed by a deity, tries tens of thousands, even millions, of different signal timing plans in the digital twin city.
- Evaluation: With each trial, the agent immediately sees the results in the model -- what is the queue length under this plan? What is the average delay? Will it cause congestion overflow in a certain direction?
- Learning and Iteration: Based on the evaluation results, the AI continuously adjusts its strategy, gradually "learning" which timing plans work best under which traffic flow patterns. It discovers that during the morning rush hour, the green light time on main roads should be extended; during off-peak hours, induction control should be used more, giving green lights only when cars are present.
- Convergence: After billions of simulations, the algorithm ultimately converges from infinite possibilities to an "optimal timing curve" for the current traffic flow state.
Dispatch Execution and Closed-Loop Feedback:
This optimal plan is dispatched in real time to the intersection's signal controllers for execution. At the same time, the intersection cameras continue to observe the actual results after execution and feed the data back to the brain. The brain compares the difference between the "prediction" and "reality," and uses this to correct its traffic flow model, forming a never-ending "learning-optimization" closed loop.
This process unfolds 24 hours a day in every corner of the city. It is not just about traffic signals, but also includes:
- Emergency Rescue Route Planning: When 120 receives an emergency call, the brain no longer simply uses a navigation app to plan a "shortest path." It instantly analyzes road conditions across the city, not only planning a "fastest path," but also pre-controlling all signal lights along the route, opening a smooth "green life channel" for the ambulance. It converges the uncertain problem of "finding a way" into the certain outcome of "creating a way."
- Bus Route and Frequency Optimization: By analyzing citizens' travel origin-destination data, the brain can identify areas with significant bus service gaps and lines with overly sparse or dense frequencies. It can simulate the effects of adjusting routes or adding or reducing trips, helping bus companies converge their limited resources into service plans that best meet citizen demand.
- Dynamic Parking Guidance: Aggregating real-time information on available parking spaces across the city and predicting future trends. When a driver heads toward the city center, the brain can proactively push a navigation instruction while they are still 2 kilometers from their destination: "The XX shopping mall parking lot ahead is expected to be full in 10 minutes. Your route has been re-planned to the YY building parking lot 500 meters away, which still has over 50 spaces."
In these scenarios, the intelligent computing center plays the role of a "city-level real-time optimizer." It uses computing power to replace the trial and error that previously required significant manpower and time, elevating the granularity of urban governance from "planning by year" and "adjusting by month" to "responding by the second," reaching an entirely new level.
5.2 Anti-Fragile Design: The "Way of Convergence" in Crisis Response
If the City Brain's everyday governance is the "superior doctor" treating disease before it occurs, then when facing extreme weather, public health events, and other sudden crises, the intelligent computing center must play the role of the "lead surgeon" in the emergency room.
The high coupling of modern cities, while improving efficiency, also makes them more "fragile." A torrential rainstorm can cause traffic paralysis, power outages, and communication failures, triggering a chain reaction.
Our goal is not merely for the city to "survive" a crisis, but to draw on Nassim Taleb's concept of "anti-fragility," enabling the city to learn, adapt, and become stronger and more resilient after experiencing shocks and chaos.
The intelligent computing center is the core engine for achieving the city's "anti-fragile" design. The key, once again, is "the art of convergence" -- rapidly converging the immense "uncertainty" brought by a crisis into "certain options" with clear probabilities and consequences for decision-makers to choose from.
5.2.1 The First Level of Convergence: From "Unknown" to "Foreknown" -- Disaster Simulation
The first step in anti-fragile design occurs before the crisis hits. The intelligent computing center must become the city's "disaster simulation laboratory."
Taking extreme rainstorm flooding as an example:
- Constructing the "Possibility Space": The operations team needs to collaborate with meteorological, hydrological, and municipal engineering departments to build a high-precision urban flooding model. This model includes not only the city's 3D topography and land cover (concrete, grass), but also the topology, diameter, and siltation of all underground storm and sewage pipe networks. Then, various extreme meteorological parameters are input to create thousands of "possible rainstorms" -- "150mm of short-duration heavy rainfall in 1 hour," "300mm of total rainfall over 3 consecutive days"...
- Computation-Driven Convergence: The intelligent computing center mobilizes massive computing power to simulate the city's response to each "possible rainstorm." The computer tells us:
- How will rainwater amass and flow on the surface?
- Which areas' drainage networks will reach saturation first?
- Which underpasses, subway entrances, and underground garages will begin to flood, and after how many minutes? To what depth?
- Which areas will experience power outages due to flooding?
- Forming "Deterministic Contingency Plans": Through this massive simulation, we converge the vague threat of "an unknown rainstorm" into a series of specific, visualized "risk maps" and "emergency response plans." We will know in advance which are the city's most dangerous flooding points, the order in which they will be submerged, and how many water pumps and rescue workers need to be pre-deployed.
This "advance simulation" based on computing power is of immeasurable value. It transforms the city's emergency response from "mending the pen after the sheep is lost" after a disaster to "preparing for a rainy day" before it strikes.
5.2.2 The Second Level of Convergence: From "Chaos" to "Order" -- Real-Time Emergency Command
When a crisis truly arrives, the intelligent computing center must instantly switch to "wartime mode," becoming the "central nervous system" of emergency command.
At this point, its task is to rapidly converge the chaotic, fragmented information coming from the scene into a clear, comprehensive battlefield situation map, and provide optimal resource deployment recommendations.
Information Convergence:
- Input: Real-time rainfall maps from weather radar, flood sensor data from across the city, photos and videos uploaded by traffic police and citizens, help requests about "flooding" on social media, power outage reports from the power company...
- Processing: The City Brain uses AI algorithms to perform second-level cleaning, cross-verification, and fusion of this multi-source information, eliminating rumors and erroneous data.
- Output: On the command center's big screen, a dynamically updated "live map" is presented, marked with real-time flood points, the locations of trapped people, interrupted roads, and available rescue forces.
Decision Convergence: Based on this "live map," the commander faces a series of difficult decisions: Which residential area should the limited rescue teams be sent to first? All three roads to the hospital are cut off -- which one should be cleared first? Should residents of a certain area be ordered to evacuate? At this point, the intelligent computing center starts its "convergence engine" once again, rapidly simulating and assessing the consequences of different decision options:
- "If we send the team to Area A first, they can rescue 500 people in an estimated 30 minutes; but Area B has lower terrain, and the water level may reach the first floor in 30 minutes, causing more severe consequences."
- "If we clear Road 1, it will take 1 hour, opening a lifeline; but if we simultaneously split our forces to clear Roads 2 and 3, although each will only be half cleared, a temporary rescue loop can be formed."
It does not make the decision for the commander. But it presents the complex, dynamically evolving consequences behind each option to the commander in a clear, quantitative way (for example, the estimated number of people rescued, the time required, and the resources consumed). It transforms the commander's decision-making from a "gambling-style" approach based on intuition and incomplete information into an "evidence-based" approach grounded in data and simulation.
5.2.3 The Manager's Challenge: Building Trust Between Human and Machine
Whether in everyday governance or crisis response, the success of the City Brain ultimately depends on one key factor: trust between humans and machines.
As managers of the intelligent computing center and decision-makers in urban governance, we must clearly recognize:
- Algorithms are not omnipotent; they also have biases and blind spots. An algorithm's decisions are highly dependent on the data it has "learned." If the historical data itself contains biases (for example, insufficient sensor density in a certain area), the algorithm's suggestions may also be wrong. Decision-makers must never blindly follow the machine's output and must maintain critical scrutiny.
- Efficiency is not the sole measure of value. An algorithm may provide a "globally optimal" traffic timing plan, but this plan might sacrifice the interests of a small number of residents on a side street. An "optimal" evacuation plan might require residents to abandon their property. At such moments, decision-makers must use human ethics, fairness, and compassion to balance, and even overrule, the algorithm's "cold" suggestions.
- Technology must be deeply integrated with governance processes. The City Brain is not an IT system that can be "bolted on." If its suggestions cannot be accepted and executed by the existing workflows of various city departments (transportation, emergency management, urban management, health), it is worthless. The operation of the intelligent computing center must proceed in tandem with the reform of urban governance systems and mechanisms.
Conclusion: Building a City That "Thinks"
Chapter 5 paints the first and most important application picture of "using well" the intelligent computing center.
Urban governance is the ultimate expression of the "art of convergence." Here, the intelligent computing center is no longer a pile of cold machines; it truly becomes a city's "brain."
- In daily life, it is a meticulously calculating "housekeeper." It converges the city's pulse into a steady rhythm of operation. Through continuous optimization of traffic, environment, and public services, it makes city life more efficient, convenient, and comfortable.
- In times of crisis, it is a calm and composed "chief of staff." It converges unknown risks into known contingency plans; it converges chaotic situations into orderly action. It uses computing power to clothe the city in a sturdy suit of armor called "anti-fragility."
"Using well" the intelligent computing center means deeply embedding this powerful engine into every capillary of urban governance. Our goal is not to create a "machine city" controlled by code, but to build an organic living entity that, driven by data and wisdom, can continuously learn, evolve, and "think."
This living entity can better serve every one of its "cells" -- every citizen living within it. This is the ultimate meaning of technological empowerment and the founding aspiration of the nation in building intelligent computing centers.