laitimes

Yuan Hui, chairman of Xiaoi Group: In the next 20 years, the biggest threat to mankind is the carbon-silicon dispute

author:The Paper

·“ Starting with opening your eyes in the morning and closing them at night, is your life and work more like a machine, or like a person? If your life and work are more like a machine, but your abilities are not as good as artificial intelligence, then what is the meaning of your existence? ”

"When the first iPhone was released in 2007, no one would say it was a watershed moment in its time. But after 3 years, we saw countless applications emerging from the App Store, and finally realized that the mobile Internet was in full swing. "Now we see an opportunity like 'Windows' emerging in the era of AI, an opportunity involving the underlying operating system."

On September 7, Yuan Hui, chairman and CEO of Xiaoi Group, said in an exclusive interview with The Paper (www.thepaper.cn) that there is Windows in the era of personal computers, iOS and Android in the era of mobile Internet, and large models in the era of AI. Because the underlying infrastructure has changed, all the applications we see now will be reshaped, whether it is social, e-commerce, or gaming. "Imagine what the house looked like when kerosene lamps were used? After connecting water and electricity, what is the shape of the house? This is the disruptive reinvention that comes with every change in the base. ”

Yuan Hui, chairman of Xiaoi Group: In the next 20 years, the biggest threat to mankind is the carbon-silicon dispute

Yuan Hui, Chairman and CEO of Xiaoi Group, delivered the "Big Language Model Bridges the Human-Machine Gap: Is Human Subjectivity Strengthened or Weakened" at the Bund Conference Insights Forum.

On the same day, Yuan Hui delivered a keynote speech at the Bund Conference Insights Forum "Big Language Models Eliminate the Human-Machine Gap: Will Human Subjectivity Strengthen or Weaken" "Will Big Models Accelerate the Arrival of AI Singularity?" , elaborates in more detail on his thinking on the big language model. He believes that large models must solve three core challenges: controllable, customizable, and deliverable. In the next 20 years, the biggest threat to mankind must be the battle between carbon-based life and silicon-based life.

"Controllable, customizable, deliverable"

On the same day, Yuan Hui delivered a keynote speech at the Bund Conference Insights Forum "Big Language Model Bridging the Human-Machine Gap: Is Human Subjectivity Strengthened or Weakened". In both this speech and interview, Yuan Hui talked about a question that is widely concerned at the moment: Will large models accelerate the arrival of the artificial intelligence singularity (machine intelligence surpasses human intelligence)?

Yuan Hui's point is clear: no. "Because the basic theoretical framework of artificial intelligence in the past 60 years has not been broken, without breaking through this framework, it is impossible in principle to produce the singularity we expect, that is, the emergence of AGI (general artificial intelligence), and human beings have created a 'new species' with self-awareness that is difficult to be positioned as a tool."

"But the prologue has already begun." Yuan Hui believes that the big model is the prelude to AGI. At present, its ability to generalize and the emergence of intelligence, including thought chains and inferences, have some basic core elements of human beings.

If the future of big models isn't general AI, what is the core value of big models? What is the biggest difference between the artificial intelligence technology of the past and the big model today?

"This difference is that in the past, AI technology could only be commercialized in certain industries and scenarios, such as call centers, smart cities, and certain specific software and hardware. But the greatest value of today's big model is its ability to drive change across all industries at the same time. This is the beginning of a truly golden decade for artificial intelligence. Yuan Hui said.

He cited the news media as an example, in the PC (personal computer) era, watching news may go to the web version of Sina's three major portals. In the era of mobile Internet, I will look at Toutiao or Douyin, or some new media based on mobile Internet. And what will the era of artificial intelligence see? "You can see every generational change from the media itself, and you know that every industry changes. So this time is a huge industrial change, and new leaders will emerge in the process. ”

In Yuan's view, this is especially important for China. "In the past few decades, it has been basically difficult for Chinese companies to enter the competition of the underlying framework of operating systems, chips, and databases. Because there are already results in the market that have been made with a lot of resources, they can be used at a low cost. But today is different, China has a new structural opportunity that has not been seen in the past 40 years, China must have its own big model, its own new operating system in the era of artificial intelligence. ”

Yuan Hui believes that in order for the big model to truly promote the progress of human science and technology and make life better, three core challenges must be solved. "The first is controllable, in terms of ethics, values, and ideology, whether they can be controlled or not, and if they are uncontrollable, they cannot be used at all; The second is customizable, if you want to use a large model, a fundamental appeal must be that you want the answer output by the large model to be the answer you want, not a serious nonsense, it must be customizable; The third is deliverable, and no one today wants to spend another billion dollars, 10 billion dollars to reinvent a big model like OpenAI, we should use a big model at a lower cost. Controllable, customizable, deliverable, it must be able to target business scenarios. ”

"Humans can be reduced to two roles"

During the speech, Yuan Hui asked the audience a question: From opening your eyes in the morning to closing your eyes at night, is your life and work more like a machine, or like a person?

"A lot of my friends have told me very candidly that their lives and work are actually more like a machine. Then the second question follows, if your life and work are more like a machine, but your abilities are not as good as artificial intelligence, then what is the meaning of your existence? Yuan Hui asked.

He believes that if this continues, human beings may be reduced to two roles. "In the first role, our bodies will be the providers of energy, because our bodies can also contribute energy; The second role, like the magazine on the cover of The New Yorker, puts a collar around your neck and humans will become providers of entertainment. ”

Yuan Hui agrees with Hawking and Musk that the biggest threat to humanity is not humans themselves, but artificial intelligence. British theoretical physicist Stephen Hawking said in an interview with the BBC in 2014: "The development of comprehensive artificial intelligence may mean the end of mankind." He argues that AI is redesigning itself at an increasing rate, while humans are limited by slow biological evolution, cannot compete, and will be replaced. Tesla CEO Elon Musk has also repeatedly warned that AI will threaten the existence of human civilization.

"The biggest challenge for mankind today is actually not the conflict between regions and countries that are currently intuitively seen, and the biggest threat to mankind is that in the next 20 years, it must be the dispute between carbon (base life) and silicon (base life)." Yuan Hui said.

The metaverse pushes humanity to rediscover the world

In September last year, Yuan Hui talked about the relationship between the metaverse and AI in the "metaverse chat room" hosted by The Paper. In this speech, he addresses this topic again.

"Many people will think that if AI is up, the meta-universe will be cold, this matter itself is a wrong cognition, the core of the meta-universe is precisely artificial intelligence, without the outbreak of artificial intelligence is impossible to create a large number of virtual worlds, relying on humans to construct this thing itself is impossible." The eve of the outbreak of the metaverse must be the outbreak of AIGC (artificial intelligence generated content), which will create a large number of people, things, and fields, and create a large number of native worlds. ”

At present, in the field of extended reality (XR, which is a collective term for virtual reality, augmented reality, mixed reality and other technologies) devices, the most concerned and anticipated manufacturer is Apple. In June this year, Apple launched the mixed reality headset, Vision Pro, which is not yet on sale. Yuan Hui believes that headsets are an entrance to the metaverse, but only part of it. "It's like using AI to do face recognition or image recognition, which is not at the same level as AI with cognitive intelligence." As far as Vision Pro itself is concerned, it does not have brain intelligence, but only takes you into the meta-universe, and what really dominates the meta-universe is cognitive intelligence. Yuan Hui said in the interview.

In Yuan Hui's eyes, the metaverse is not some new way for humans to entertain, but a way to promote humans to re-understand the real world. In the process of mirroring the real world, the real world merges with the world created by AIGC to create a world where reality and fiction coexist, constantly pushing human beings to re-understand themselves. According to the current trend, if human beings cannot re-understand the world and themselves, artificial intelligence may destroy human beings.