Chapter 349: China Trip, We Can Do It Too!
The sudden emergence of Deep Red is not only a major commercial victory, but commercially, Tencent’s stock price rose 100% in just five days. Although it is still some distance from the peak, this is already very impressive.
You know, Tencent’s market capitalization is huge, and even with its major shareholder Naspers continuously reducing its holdings, it still achieved such a rise. This shows the power of Deep Red when such an elephant dances.
It is also a major victory at the China technology level. Globally, it is another case of America can do it, China can do it too; America can do it, but why can’t we?
Especially in Europe, old Europe has rarely seen a wave of reflection.
Marcin Lewandowski is a German blogger, mainly doing podcasts, also on YouTube, and has always focused on the artificial intelligence field, making him one of Europe’s top bloggers in this field.
On March 25th, he pushed his latest podcast episode on various social platforms: “Marcin Lewandowski: Why Europe’s AI Has No Future – In Conversation with French Technology Director Clara Chapaz”
Here is the specific program content:
“Welcome to the latest episode of EU-Startups, I’m your old friend Marcin. Today we have invited a heavyweight guest to have an in-depth chat about GPT, LLM, the development of the artificial intelligence industry, China’s newly launched Deep Red, and Europe’s future in the artificial intelligence field.
She is the ministerial-level representative in the French artificial intelligence and digital technology field, Clara Chapaz.”
They conducted the interview via remote connection. Watching the video alone, it appears somewhat rudimentary, with everyone at home connecting remotely via Zoom meeting.
But after all, Marcin’s main focus is podcasts, and podcast audiences mostly listen rather than watch, so the video aspect is not that important.
“Hello everyone, I’m Clara. My job is mainly to ensure France’s leading advantage in the artificial intelligence field and guide companies to invest in artificial intelligence and digital technology in France.”
Ministerial-level representative, this position is actually already very high.
But Marcin does not believe that the other party really understands artificial intelligence or can truly lead France to become a leader in the artificial intelligence field.
So he took the initiative to challenge: “Clara, do you think France is currently a leader in the artificial intelligence field?
But we haven’t seen France produce any outstanding achievements. America has GPT, China has Deep Red, and France seems to have nothing in this regard.”
Before Marcin could finish, Clara interrupted: “Of course, we are of course a leading country, and China and America are also leading countries.
There is not just one leading country.
France has a long history in the artificial intelligence field, with a large amount of technical reserves, and we also have a series of startups.
Including the French government sponsoring some artificial intelligence projects, like Lucie, Mistral AI, etc. We believe France will soon produce results that shock the world, like Deep Red.”
Indeed, Lucie did shock the world to a certain extent.
Clara’s logic is that it’s not only the top two that are leaders; as long as you’re above the median, you can call yourself a leader.
With 197 sovereign countries globally, France can at least rank in the top five, so how can it not be considered leading?
Marcin continued: “Can it really shock the world?”
Clara said firmly, “Of course. The French government has unprecedented support for artificial intelligence. We have the best talents in the artificial intelligence field. France has always had an absolute advantage in the mathematics field.
As we all know, the essence of artificial intelligence is mathematics. France is the mathematics center, and we will soon convert our advantage in mathematics into an advantage in artificial intelligence.”
This made Marcin, a German, very unhappy. If Göttingen hadn’t declined, how would it be Paris’s turn to be the mathematics center?
“That’s right, Clara. Unlike China, which is restricted by computing power, so why is China’s artificial intelligence progress faster than ours?
I bought a Deep Red account from a Chinese seller on eBay and experienced it personally. First, it’s free, and second, its effect is even better than the paid GPT-4.
I was shocked at how Chinese people achieved such good results using restricted Nvidia computing cards.
But we can’t.”
Marcin’s question treats Germany and France as a whole, all under the big framework of the European Union.
Overseas, many tech enthusiasts want to try Deep Red. Self-media bloggers get accounts from Chinese peers through networks and spread them widely, while ordinary people buy them on Amazon and eBay.
China only restricts registration with Chinese mobile phone numbers, not IP addresses, and does not prohibit overseas IPs from logging in.
So many Chinese people made a fortune selling Deep Red accounts, with individual accounts ranging from 50 to 100 US dollars.
Clara explained: “The most important reason is that the European Union has many regulations on artificial intelligence, with many compliance requirements to avoid risks that artificial intelligence may bring. We need to develop harmless AI and avoid ethical issues brought by AI.
Obviously, China’s artificial intelligence doesn’t need to consider so much.
Moreover, China has a large number of engineers returning from Silicon Valley. They may have learned the engineering implementation logic of GPT-3 and subsequent GPT-4 through some channels, while we have to rely entirely on ourselves.”
This French bureaucrat’s logic boils down to two points: we have higher and stricter requirements, and China can copy American technology.
Marcin was speechless. He really couldn’t understand why this woman could become a ministerial-level senior official in charge of artificial intelligence and digital technology.
This answer shows no level at all.
“Are you saying China plagiarized American technology? And as a result, their plagiarism turned out better technically?” Marcin asked.
“I’m just saying it’s possible, not that it definitely is.” Clara backtracked: “Like in the moon landing, they used American Apollo moon landing technology, but they did it better. They even landed at the lunar south pole, something the Apollo moon landing couldn’t do.
Replicating the same miracle in the artificial intelligence field is not impossible.”
Marcin immediately retorted: “Logically possible, but China only took two months. Two months after GPT-3 was released, they launched Deep Red, and before GPT-3, OpenAI was just an insignificant small company.
I think plagiarism is almost impossible.
We shouldn’t dwell on how China did it; we should discuss how Europe can catch up.
You mentioned earlier that we have advantages in talent and technical reserves, but aren’t we too focused on laws and regulations?
I have a very clear feeling: we have the European Data Protection Supervisor, the EU AI Act, compliance and risk management governance platform, EU AI regulation expert team, European Policy Center.
These institutions can all intervene in AI research and development, discussing how to ensure AI has no risks or ethical issues.
But we don’t have artificial intelligence yet? We don’t have artificial intelligence, yet we’re spending massive manpower and resources on AI regulation?
I’ve observed that podcasts in the same artificial intelligence field in Silicon Valley discuss exciting new technologies, which teams made them, and how these technologies can change our work and life.
While European artificial intelligence podcasts discuss European data privacy, AI regulations, and AI privacy challenges in Europe.
Clara, as France’s ministerial-level representative, what do you think of this issue?”
Marcin found this ridiculous: Europe has no artificial intelligence yet, but keeps discussing how to govern it.
“Because AI risks far exceed imagination, we need to balance risks and efficiency,” Clara said.
Clara had sensed the other’s unfriendly attitude, but she fully understood that podcasts with conflict and confrontation attract audiences. Since she chose to participate, she was prepared for this.
“Shouldn’t we solve problems after they appear?
Instead of discussing potential problems before they appear, and how to solve problems A, B, C, discussing a bunch of issues.
What if none of them appear in the end? We’d have to discuss again and come up with proposals.
Isn’t this doing useless work?
In other words, if we had artificial intelligence comparable to Deep Red and GPT-3 today, but it doesn’t comply with EU artificial intelligence laws and regulations, would we prohibit it from providing services to the public?” Marcin asked.
Marcin’s confusion was something Clara knew how to answer, and why the EU does what it does.
Simply put, it’s path dependence. Europe has no technology giants; the money is all taken by Google, META, Amazon, Microsoft. The digital taxes they pay to Europe are minimal, so Europe can only obtain tax revenue through fines.
Nominally huge fines for violating laws and regulations, but in reality, it’s just another form of tax.
Similarly, why does Europe study so many AI regulation laws? For the same reason: to fine you Silicon Valley AI giants. I don’t have such companies myself, so I rely on fines for tax revenue.
So it’s not researching before problems appear; it’s because GPT has already appeared and has a large number of users in Europe that we’re studying how to fine them, at a balance point acceptable to both sides.
For politicians, this can be done but not said.
Outsiders think Europe’s politicians are a bunch of idiots, but it’s path dependence; researching fines is so direct.
Nurturing native technology giants? Can old Europe do it? It failed in the PC and mobile internet era; can it succeed in the artificial intelligence era? Clara doesn’t believe it herself.
But on the surface, it needs a passable reason: “We need to guide the future development direction of global artificial intelligence and reveal potential risks of artificial intelligence.”
This is what politicians are like; this sounds so nice.
“As for your second question, we can grant additional exemptions to such native European companies.”
The entire interview ended hastily. Once exemptions were mentioned, Marcin lost interest in continuing.
If exemptions are possible, then why set up so many institutions and use so much government tax revenue to support so-called artificial intelligence regulation experts?
After Clara left Zoom, Marcin unleashed full firepower:
“See? This is our official in charge of artificial intelligence and digital technology. Let me add two points: she does have a computer technology background, but Clara Chapaz was previously CTO at LV and VC. The latter is a second-hand luxury goods online sales platform, neither of which has anything to do with artificial intelligence technology.
Her understanding of artificial intelligence is even worse than that of an undergraduate student majoring in computer science.
I’m not saying she’s unqualified; I mean across Europe, every one of these bureaucrats is unqualified in the artificial intelligence field. We don’t know to poach from Silicon Valley. China has already proven in the PC and mobile internet era how useful people returning from Silicon Valley are.
Why don’t we do it? There are plenty of Europeans in Silicon Valley; why don’t we find scientists in the artificial intelligence field among them, invest in them, and let them return to start businesses?
This is what government departments should do, not research damn regulations and damn risks.”
Marcin went all out, determined even if it meant he could no longer interview government officials.
“We’ve wasted too much time. The past is the past, but now artificial intelligence shows unprecedented potential. All giants are investing in the LLM direction, and Europe is obviously lagging behind again.
Indeed, as she said, it’s not just first and second that count as leaders. The problem is, now the big data-driven LLM is a clear winner-takes-all game.
With America led by OpenAI, and Google, META seizing the market, where do European companies find markets? Go to China? Or Africa?
France’s existing mathematicians may have some better than Professor Lin with more outstanding achievements, but they are old. Can they really switch to the artificial intelligence field and produce outstanding results?
As for young scholars, can they compete with Professor Lin?
Sorry, I’m very pessimistic about Europe’s artificial intelligence future, not just because we’re behind now, but because we have such bureaucrats deciding our artificial intelligence future, which makes me even more pessimistic.
Here, I need to conclude early. Although the LLM era has only begun for half a year, I’ve already seen that Europe’s AI has no future. Thank you for watching.”
Americans and Europeans are both in the Europe and America camp, the Western camp, but their mindsets toward Deep Red are completely different.
Americans feel excited that a competitor has finally appeared. Nasdaq US stocks are falling, secondary market performance is pessimistic, but the primary market is very active.
The emergence of Deep Red means that in this field, in the artificial intelligence field, the dust has not settled, and everyone has a chance.
If China can produce Deep Red, why can’t Silicon Valley produce the next Deep Red?
Silicon Valley venture capital institutions began frequently visiting Silicon Valley artificial intelligence startups, preferably all Chinese descent, mostly Chinese descent second best, and AI companies with no Chinese descent at all are ignored.
Nilanjan was poached by META. Nilanjan at Stony Brook University for so many years, every PhD he trained received offers from big companies like Google, META, Microsoft, with annual salaries of millions of US dollars.
Or start their own company. If your artificial intelligence startup has students of Professor Nilanjan, investment institutions will naturally look favorably on you, reducing the difficulty of raising funds by 90%.
Those students who thought Nilanjan had disappeared and switched mentors, thinking they had it made for life, received big employment gift packages. Those not yet graduated have been pre-booked by technology giants, and graduates ate a wave of bonuses.
Of course, those benefiting are not just them; Lin Ran’s math class students too.
They were admitted to Lin Ran’s math class in the second half of 2020, total 20 people, now juniors in their second semester, about to face graduation destinations.
20 people, Shanghai Jiaotong University provides 20 guaranteed postgraduate recommendation spots. Simply put, as long as you don’t fail courses and pass level 6, you can get guaranteed recommendation; where to go depends on your ability.
With such good treatment, plus Lin Ran’s reputation, Lin Ran’s math class has almost become Jiaotong University’s trump card.
In these two years, quite a few students who could get into Tsinghua or Peking chose Shanghai Jiaotong University just for Lin Ran’s math class.
This is still based on the fact that Lin Ran’s math class can only be entered after enrollment.
If high school entrance could directly go to Lin Ran’s math class, many top scorers would be willing to come to Jiaotong University, because they expressed similar willingness during application consulting: if I can ensure entry to Lin Ran’s math class, I’ll come.
On March 31, Thursday, in a meeting room at Apollo Technology, students from 3 cohorts of different grades in Lin Ran’s math class met Lin Ran.
“Classmates, really sorry, I have too many things, can’t guarantee weekly classes for everyone. This should be the first class this semester.
Before class, let’s chat a bit. I know many of you are interested in artificial intelligence. Not everyone wants to do theoretical mathematics research, treating Lin Ran’s math class as a springboard for future employment positions.
Teacher Li from the school also told me that some asked him if they could intern at Deep Red.
Deep Red is already online, so I thought today is as good as any. I invited everyone here for class and to answer questions.
Wanting to intern at Deep Red is no problem. All 60 classmates here can come.
Initially, I wanted to set a slight threshold, like top 50% in professional course GPA, just half.
But then I thought, we aim for interest-oriented. This is a pure mathematics class; high scores in theoretical mathematics don’t mean equal knowledge accumulation in artificial intelligence. After all, not everyone is like me, an tireless learning machine.
So I decided against it. As long as you want, apply and you can intern at Deep Red.
No ability threshold; the ability threshold was crossed when entering Lin Ran’s math class. The threshold here is mainly security.
Including technical security and my personal safety. This threshold is very low; I believe everyone has no issue.
So welcome everyone to intern at Deep Red this summer. You can experience how the latest, most cutting-edge, top artificial intelligence company works and operates.”
After Lin Ran finished, the audience erupted in thunderous applause, almost lifting the ceiling.
Everyone knew what this meant.
Simply put, this resume equals a direct PhD admission letter from any university in the world, equals an offer from any company in the world.
The top welfare among top welfares.
“Okay, everyone calm down.” Lin Ran smiled; he was happy to give young people opportunities.
Especially these young people here, his direct direct lineage.
“Additionally, I plan to recruit three PhD students: one AI PhD, one pure mathematics PhD, and one applied mathematics PhD in the aerospace field.
This is for the juniors, since you’ll face destinations next year.
I believe my PhD students will have the world’s top resources, so it will be selecting the best of the best. Not necessarily from you here; you’ll compete with global undergraduates and master’s students.
If interested, prepare in advance,” Lin Ran said.
He has long been a PhD supervisor, just hasn’t recruited any.
Being a PhD supervisor is a qualification, not a requirement to recruit every year.
In the past, on the Chinese Internet, people kept asking if Lin Ran recruits PhDs. Countless students wanted to apply, from this school to others, undergraduates to master’s.
Even some who already completed a PhD wanted to do another under Lin Ran.
A freshman from Lin Ran’s math class raised her hand; Lin Ran nodded to her.
She stood up and asked: “Professor Lin, hello. I’m Li Siqin, a freshman in Lin Ran’s math class. I want to ask, are the three PhD spots ongoing every year, or just this year?”
Lin Ran explained: “Just this year. Later depends on the situation, probably after training this cohort before recruiting more. My energy is limited.”
PhD supervisors have limited annual PhD spots; can’t recruit infinitely. But that depends on the person; academicians have nearly unlimited spots, let alone Lin Ran, such a Two Houses academician.
Li Siqin continued: “Then isn’t this unfair to us freshmen and sophomores?”
Lin Ran nodded faintly: “Of course. Nothing is fair.
You need to adapt early. On providing Deep Red internships, it’s equal treatment, everyone can come; that’s fair.
On recruiting PhDs, juniors get priority to apply; that’s unfair.
The world is like that.”
“Professor Lin, since the first thing can be fair, why can’t the second?” Li Siqin was indignant, eyes slightly red, feeling she lost a once-in-a-lifetime opportunity.
Other classmates whispered among themselves.
Some surprised at her boldness, daring to be so aggressive to Lin Ran.
Others thought she was fighting for everyone’s rights.
