After the meeting, the group stayed in Xini for a few more days, which was a rare moment of relaxation.
Unfortunately, happy times are always short-lived. Except for Meng Fanqi and Tang Juan, who have dropped out of school, the other four are still studying.
It is quite difficult to be able to come with Meng Fanqi during normal class time this time.
If I delay for a few more days, I will eventually have to rush back to school, otherwise it is not impossible to fail the third and fourth subjects.
Meng Fanqi is willing to help those around him, but he doesn't want to escort them all the way. It's unrealistic for people to finish their undergraduate studies.
The main purpose of bringing them here this time is to let them see the current situation and potential of the AI industry.
Liu Xu, Mai Mang, Qin Hui and Meng Fanqi naturally had their own arrangements and plans as they planned to stay in the country.
The most important thing he hopes a few people can help do is to build a high-quality Chinese database in the country.
This thing is the most critical factor for the success of Chinese ChatGPT in the future.
Meng Fanqi is confident that he can capture all the resources of the Western AI community with his own abilities, but he cannot solve the fundamental problem of insufficient high-quality Chinese data.
This was also the biggest reason that troubled many technology companies in China before the rebirth. High-quality Chinese data was far less than English.
And high-quality data is the oil in the AI era.
On the one hand, China's Internet started late, and its data accumulation is relatively poor. On the other hand, everyone is busy making quick money.
Who is willing to do this kind of dirty work of throwing money in, collecting data, and labeling the data?
In order for Chinese and Chinese AI to keep up with the pace in the ChatGPT era four or five years later, this matter can only be planned in advance by Meng Fanqi.
This is a big project that swallows money like a black hole. The cost of using an employee is at least 100,000 yuan a year. If you hire hundreds of people, it will be tens of millions in a year.
All you get is a bunch of ruthless data, and there won't be any revenue from this thing in the short term.
In the long term of four to five years, Meng Fanqi said that it would cost three to five billion to invest. If you want to improve the quality and scale of data, it is not impossible to invest more than one billion.
"Fortunately, we signed a sharing contract with Google, otherwise I really don't know where to get the money."
Meng Fanqi felt a little pressure. After carefully calculating a few accounts, he realized that the cost of what he wanted to do seemed to be much higher than he imagined.
An update to Google's recommended advertising algorithm is on the agenda as soon as possible.
"Liu Xu, Mai Mang, Hui Hui, the situation is probably like this. I will contact you in detail in three or four months. You are first responsible for helping me with data collection in China."
At the airport, Meng Fanqi and Tang Juan were going to Shanghai, while the others were going back to Yanjing.
After saying goodbye this time, there won't be many opportunities to meet in the future, so Meng Fanqi explained the general plan for the future, "For this data collection task, you don't need to consider revenue, so don't be stressed. This company's departments have been built It’s not just to make a penny, other departments will do part of the revenue, I will bear all the expenses, and your salary will definitely not be reduced.”
Let these classmates help him take care of the data collection first. Meng Fanqi signed a big deal with Google, and Baidu's share will not be transferred to the United States.
Just stay in Yanjing and invest in building the Chinese data infrastructure. Don't make a drawing and have to convert it into English, making jokes like squirrels and mandarins, but they are confident.
“Let’s build the data infrastructure first, and I’ll talk to you in detail when there are other opportunities in the future.”
This AI company will definitely have some cooperation with Baidu, and Meng Fanqi plans to use his connections to get more data from Baidu.
After a few of them have some experience, they can follow the people in the technical department and go back and forth with Baidu to practice. Next year, it will not be too late to let them take on a small part of other opportunities and tasks for technical cooperation with major domestic manufacturers.
Of course, Meng Fanqi will definitely be fully responsible for the technical aspects.
The main reason is that it will be more secure if someone familiar with the data is responsible for taking care of it.
"Then... come to Stanford for exchanges if you have the opportunity." Han Ci didn't give Li Feifei a definite answer a few days ago, and Meng Fanqi didn't know what she was thinking.
It stands to reason that academicians from Yenching University and Hubei University are well-known in China, but Han Ci is very clear that doing AI and statistics in a circle that should be counted is definitely not following the footsteps of Meng Fanqi, a leader in the new era of AI. AI theory and AI interpretability are cost-effective.
Just for the spectacle shown two days ago, Meng Fanqi's actual performance breakthrough was not achieved first. This mathematical proof is absolutely ignored and not available.
However, her thoughts were very confused at the moment and she had not figured it out yet.
Meng Fanqi didn't say much. He believed and respected Han Ci's own thoughts.
Not long after, he and Don Juan boarded the flight back home.
During this trip, apart from Meng Fanqi and Han Ci, who have become famous in the AI academic community, Tang Juan gained the most among the other four people. The remaining three were just strolling around and seeing the world.
As the sponsor of Meng Fanqi's experimental equipment during the competition at that time, and as the person who guided Meng Fanqi on the path of AI in his previous life, although he was a little confused, Tang Juan was still qualified to have his name on the participating team.
This also won him a ticket to Stanford. Even without the full scholarship and direct doctorate, the quality of Stanford was far better than the schools he originally planned to apply for.
If you have a friend who knows everything about you, apart from anything else, you can safely leave many inconvenient things to him in the future.
Besides, Don Juan is quite talented in many aspects, or it can be said that they are not considered talents.
For example, being familiar with the second dimension does not seem to be an advantage at any time.
Apart from this special trip to Shanghai to harvest rice and swindle a wave of shares, Meng Fanqi really couldn't think of any other use for this attribute of Don Juan.
Although Mi Huyou still has no signs of success at this time, this studio with only a few people has actually been established for nearly three years.
At the beginning of 2011, a few students who were still studying in a dormitory of Shanghai Jiaotong University established Mihuyou Studio.
No one could have imagined that this two-dimensional game company would later be able to directly reach a valuation of 200 billion yuan with two or three games.
Among all the Internet companies that have taken off in the new century, Mihuyou is unique in more than just this.
The most legendary part is that this company has only received one financing from beginning to end.
Everyone often sees news like this, where a company received Series B financing and was valued at tens of billions.
The reason for financing is actually the lack of money to do things. Just like the three founders of Mihuyou who receive a monthly salary of 4,000 yuan.
Angel round, A round, B round, C round, D round financing actually refers to the order of financing.
The further you go, the more mature the company's products, profit methods, scale, etc. become, and the risks become smaller. To obtain the same shares, more capital needs to be invested.
For ordinary companies that want to go public on a large scale, it is very common for them to raise four or five rounds of financing. Even a financing scale of several hundred million is not considered large.
Mi Huyou spent more than a year from the end of 2011 to 2012, but failed to get a single investment.
Since the three founders are all members of the Shanghai Jiaotong University's academic class, many investors have actually been interested in Mihuyou.
It's a pity that two-dimensional culture can hardly be regarded as the mainstream of the market today, let alone at the time.
In particular, the concentration of the three founders is too high, and the ingredients are very problematic.
