Self-justification

"The technical details of the food delivery algorithm? Tell me about them." Professor Zhang leaned back on the sofa, looking a bit tired.

"I'll send you the document."

"No need. Documents involving algorithm details should not be stored on your phone if possible." Professor Zhang rubbed his temples, "I don't have the energy to read any documents right now; just give me a general overview."

Lin Yuan had meticulously refined the AI Feng Shui model before, even running code validation on Hao Tuan Company's machines. Yet, in the end, his mentor didn't even want to look at it.

It seemed his mentor wasn't very interested in the technical details of the algorithm; perhaps he had just mentioned it casually out of curiosity about technology.

"After I graduated last year, I delivered food for over half a year. At that time, I was thinking that current food delivery and routing algorithms had too many shortcomings."

"After much thought, I felt that a large internet company like Good Group must not lack computing power resources."

"Even if computing power resources are expensive, it's only a one-time investment."

"Moreover, internet companies love to tell stories by grasping 'models.' Algorithm iteration involves the update of models, so it's impossible for them not to be willing to invest resources into it."

"So, the current delivery problems in the food delivery industry can only be solved by throwing resources at them."

"Then, why can't they be solved even with resources? I started from the most basic elements of AI training."

Professor Zhang listened to Lin Yuan's eloquent explanation, and his technical intuition told him that after the preceding groundwork, Lin Yuan was about to unveil the answer. So, Professor Zhang subconsciously stopped rubbing his temples.

Lin Yuan continued, "The basic elements of AI training are nothing more than: data collection, model building, and computing power investment."

"Based on my experience working at Good Group, they absolutely have the financial means to acquire enough computing power equipment. So, computing power investment isn't a problem. As for model building, it's essentially various fine-tunings based on deep neural networks. The most advanced theoretical papers in the industry are publicly available to everyone, so this point is essentially an investment in human resources. In essence, it's like investing in computing power equipment; it's all about throwing money at it."

"Since that's the case, Good Group doesn't seem to lack money at present."

"Therefore, I focused my attention on 'data collection.'"

"Data collection? What exactly do you mean? Is it that the food delivery platform collects insufficient data, or that the collected data is incomplete?"

"The data they collect is incomplete. Teacher, think about it, isn't the food delivery industry just a model where merchants and delivery riderss work together through the food delivery platform to serve diners? The most crucial factors here are all time-related factors."

"The time it takes for a delivery riders to go from one place to another, the time it takes for a merchant to prepare a meal."

"But because these time-related factors are difficult to collect, they can only resort to experience and regulations."

"For example: the stipulated delivery time for an order, the stipulated meal preparation time for a dish."

"These types of times should be flexible, but in reality, they become crude empirical estimates."

"This leads to delivery riderss running red lights, and delivery riderss and merchants having conflicts over meal preparation speed."

"And these variable factors related to time are actually completely influenced by the world we live in."

"The quality of the weather on that day and at that time, the quality of local practitioners, even the local customs, and the local economic development level. All of these will affect the final time factor."

"And collecting these things is very difficult; it's simply impossible to achieve with the data collection methods commonly used in today's AI field."

"So, I changed my approach. I don't directly collect these parameters because I simply don't know how to collect them."

"Just like in ancient times, people would see doctors when they got sick. Those old Chinese medicine practitioners in ancient times didn't know modern medical knowledge, nor did they have any modern medical testing instruments to assist them, but this didn't affect their ability to treat people."

"Our ancestors came up with another method—building their own theoretical system. That is, the theoretical system of traditional Chinese medicine."

"And this theoretical system of traditional Chinese medicine is actually an intuitive perception of the world we live in. 'Damp-heat, wind-cold,' 'using form to supplement form,' and so on, are all the most intuitive perceptions of this world. Although such theories cannot be included in the scope of modern science, from a practical perspective, they are very useful in certain aspects."

Professor Zhang listened, gradually becoming engrossed, and had now stood up, "You mean, using the mindset of traditional Chinese medicine to perform a different kind of data collection for the food delivery industry?"

"Exactly. Traditional Chinese medicine is inextricably linked to traditional Yin-Yang and Five Elements of Feng Shui. It emphasizes the balance of Yin and Yang and the harmony of the Five Elements, and the harmonious coexistence of humans and their surrounding environment. So, I thought of doing data collection from the perspective of Feng Shui."

"Feng Shui?" Professor Zhang almost dropped his jaw, "You're introducing ancient fortune-telling and divination into the cutting-edge AI field of modern IT technology?"

Lin Yuan said indifferently, "Many people say that AI training is just computer fortune-telling. It's the same principle."

"Ah?" Professor Zhang was stunned for a moment, then immediately relieved, "Hahaha~~~, what a 'same principle.' If you explain it that way, it does make sense. When humans face extremely complex things, they always try to understand them with simple models. Fortune-telling and divination are one way, and AI is also a way."

"But AI, at least, is supported by mathematical formulas, even though the process by which these formulas work is too complex and beyond the scope of human cognitive ability. But fortune-telling and divination don't have mathematical formulas, do they?"

"But, Teacher, the mutual influence between humans and the environment is constantly happening. For example, how much impact will our act of discharging sewage into a river have on us? A single discharge might still be evaluated with mathematical formulas, but what about our every breath?"

"I step on a small blade of grass while walking, I release a fish, I see a beautiful scene, I plant a plant."

"How can such tiny, repeated interactions between humans and the environment be described by mathematical formulas? At least, current mathematics has not developed to that extent. But such mutual influences are happening every moment."

Professor Zhang pondered for a moment, "You're getting philosophical here."

"Uh, I'm trying my best not to touch something as abstruse as philosophy."

"Then what does your current model look like?"

"Of course, I don't have the ability to use one model to cover such a complex relationship between humans and nature. So I changed my way of thinking—since it's too complex to unravel, why not simplify it to the extreme? Similar to the idea of Tai Chi Yin and Yang, one begets two, two begets three, and three begets all things."

"I embedded the communication between humans and the natural environment into the language system of Feng Shui, and Feng Shui emphasizes the three talents: Heaven, Earth, and Man. I then forcibly adapted this idea, using three major categories of parameters: Heavenly Timing, Earthly Advantage, and Human Harmony, to summarize it."