Becoming Rich with Daily Scavenging APP

"I think today's AI is both a very distant and very close industry. Distant, because it might represent the future of humanity.

Close, because in recent years, smart AI has achieved numerous application functions through AI painting, document processing, searching, video synthesis, and more.

Take AI painting for example: when it first appeared, people said AI-generated art was obviously fake.

But within a few years, many gaming companies fired their art departments and started using AI for painting 2D images.

When used well, players can hardly distinguish which artworks are AI-generated and which are hand-crafted."

"So, do you have a profitability demand for the AI industry?" Wen Liangfeng inquired.

"Of course not," Chen Yiyang responded decisively, "The current AI industry is still in the cultivation phase, far from reaching profitability."

Chen Yiyang went on, citing Tengxu as an example, "Tengxu created a chat social software, but for a long time it couldn't be profitable and was even forced to the brink of selling the company.

Tengxu tried many things in the meantime, like charging for registration, but they all failed.

Eventually, Tengxu relied on game agency to secure sufficient funds. Their success was actually due to using the social software to channel traffic.

I think the AI industry is the same, and won't rely on selling memberships for profits but some unexpected channels.

As long as we remain in the top tier of the AI field, many profitable methods will naturally emerge."

Chen Yiyang's explanation clearly satisfied Wen Liangfeng.

He was worried about meeting those investors who, although willing to invest, demanded short-term profitability.

If an AI software can only profit by selling memberships, how much of a failure must it be?

Setting up memberships creates an artificial barrier, keeping ordinary people out.

Ordinary people's sheer numbers actually represent huge wealth.

Excluding them is equivalent to holding the future back.

"If our AI model training succeeds, we would decide to open-source, inviting more people to join us and help improve our AI model, do you agree?"

Wen Liangfeng continued to ask.

"Of course I agree, isn't that how Google rose to power? Start by open-sourcing to attract countless developers to improve the system.

Then create their own Google suite to charge for it. Although not as profitable as Apple, they still earn amply."

Wen Liangfeng was even more pleased with Chen Yiyang's answer.

Such knowledgeable investors who are not eager for profits are indeed rare.

Rare, aren't they?

Most investments essentially go to waste, possibly returning profits only once in a thousand attempts.

Under such circumstances, some investors crave instant success, hoping to recover their costs quickly, thus interfering with the entrepreneurial team's ideas and practices.

This is human nature after all.

Chen Yiyang already determined that the thirty-something young man before him could achieve success in the future.

Naturally, he would speak words the founder likes to hear.

"I accept your investment in our team, now let me talk about how much money we need."

Wen Liangfeng felt that Chen Yiyang had passed the test of two questions.

Now it was feasible to discuss specific investment amounts.

"Name your figure," Chen Yiyang was ready for the other side to propose an astronomical investment amount of several billion dollars.

"We've already calculated the costs before. According to our calculations, to complete our own AI model, it's estimated to require five million to six million USD."

"Five hundred billion to six hundred billion?" Chen Yiyang initially misheard.

Then he suddenly reacted to what the other person had just said.

"Wait, did you say only need five million to six million dollars?!"

Chen Yiyang incredulously asked again.

"Yes," Wen Liangfeng nodded, "This is the conclusion our team reached after precise calculations."

"Are you sure you haven't mixed up units?" Chen Yiyang still didn't believe the price quoted.

"No mistake, definitely between five million to six million dollars," Wen Liangfeng reaffirmed with a nod.

Honestly.

If this investment wasn't confirmed through inside information.

Chen Yiyang would really have thought the other was a scammer and turned to leave.

Five million to six million dollars, what kind of concept is that?

It's less than fifty million RMB!

Mind you, training the ChatGPT-4 model alone burned through several billion dollars.

That's just the cost of training the model, not including team salaries.

And Google's AI division spent as much as 1.9 billion dollars just on model training in one year.

In contrast.

Wen Liangfeng said their team needed only five to six million dollars to train an AI model.

It's even more cost-effective than Pinduoduo.

Even Pinduoduo never claimed their product prices were equivalent to other platforms' shipping costs.

Five million dollars for Tengxu or Qian Du, probably wouldn't even cover poaching fees.

Those industry elites often require tens of millions in breach penalties to be poached.

A team should have five or six top talents and dozens to hundreds of core personnel, right?

The salary costs for these talents would probably exceed tens of millions per month.

And now, this future top AI model team's founder claims he needs only five to six million dollars to train a top AI.

"Are you sure, only five to six million dollars?" Chen Yiyang still couldn't help but question suspiciously.

"You think this expense is too low, right?" Wen Liangfeng adjusted his glasses and continued, "Of course, this is just the cost of training an AI model.

If we want to ensure long-term competitiveness, consistent funding is required.

After training the model, roughly 200 million to 300 million RMB would be needed annually."

"USD?"

"RMB."

Chen Yiyang did some calculations.

He still felt it was too cheap.

If this price were reported to those major Internet companies, the leadership there would likely think their subordinates had gone mad.

Two or three hundred million per year for AI?

Do they think my Honor of Kings can't sell skins, or Peacekeeper Elite has no players?

Who are they underestimating?

"Here's the thing," Wen Liangfeng explained to Chen Yiyang, "We think many AI models on the market are somewhat bloated, and the training methods are quite wasteful and complex.

So we aim to improve training methods, such as sparse training and dynamic network structures, to reduce model parameter volume while maintaining performance, thus decreasing computational dependency.

This would significantly lower costs. However, our idea hasn't been accepted by investors domestically.

They think our price is too low, suspecting we're planning to take the money and run.

That's why I came to California to visit some high-tech companies. But unfortunately, leadership in companies in the United States also believes this. The CEO of OpenAI even said a company with only tens of millions of dollars in budget a year cannot compete with them in the AI field."