MBZUAI President Xing Bo wrote an article: The "renaissance" of artificial intelligence has arrived

Author: Xing Bo

Source: The power of the machine

Image source: Generated by Unbounded AI

On the issue of AI risks, the bigwigs from all walks of life are also disagreeing, and a large-scale debate broke out on social media a while ago.

Some led the way in signing a joint letter calling for an immediate moratorium on AI labs and for the government to tighten regulation. There are also many people who are very positive and optimistic about the development of AI, believing that the development of AI is far from posing a threat to mankind, and regulation will only degenerate into an umbrella for the interests of a few giants.

The change in the UK government's attitude towards AI technology before and after is an example of the "threat theory" that is so loud. In March, the UK government released a white paper promising not to stifle AI innovation. Just two months later, the "AI apocalypse" was rampant, and the government began to talk about putting guardrails on AI and urging the U.S. government to do the same.

It is against this backdrop of public opinion that Xing Bo, president of Mohammed Bin Zayed University of Artificial Intelligence and professor at Carnegie Mellon University's School of Computer Science, expressed his views.

As one of the world's top computer science professors, Xing Bo has been outspoken in his criticism of these "threat theories" and "doomsday theories", "creating a public opinion full of pessimism and exacerbating public fear and anxiety." But there is a "great gulf" between the reasoning and the conclusions of these claims. He pointed out that those who create and advocate pessimism "do not insist on the rational analysis and rigorous argumentation that a member of an educated society should have." 」

Xing Bo is positive and open to the development of AI. He believes that AI is ushering in a "renaissance" of the 21st century, and it will also fundamentally change the way humans acquire knowledge and solve problems, bringing us into an age of empowerment that will empower humanity what it cannot do and promote human well-being.

This article is a Chinese translation of Xing Bo's contribution (English), compiled by Machine Heart, and the following is the full text.

• Artificial intelligence is fundamentally changing the way we acquire knowledge and solve problems. *

Before chatbots and image generators made their high-profile debuts, AI was already quietly integrated into people's daily lives. It can recognize faces, open your phone, and act as a translator and guide as you travel. When you want to relax at the end of the day, it can even pick a movie for you.

But the chatbot revolution has been accompanied by some pessimistic warnings that equate the growing capabilities of AI to existential threats such as nuclear or natural disasters. Some opinion leaders on the Internet say that there will be an all-knowing and all-powerful artificial intelligence ghost in the future, and they have also created some other abstract but absurd claims. Then some of the biggest figures in academia and business have chosen to amplify these narratives, creating a pessimistic public opinion that fuels the public's fear and anxiety and fails to insist on the rational analysis and rigorous argumentation that an educated member of society should have. The voices of true researchers and innovators at the forefront of today's science are drowned out by these arguments.

A closer look at the actual threats reveals that the alleged dangers of AI are exaggerated. The indelible scars left by nuclear weapons in Hiroshima and Nagasaki, the ravages of pandemics such as Covid-19, and the melting of glaciers due to climate change are all stark images of real dangers.

The pessimistic depiction of AI is more sensationalist than scientifically realistic. Unlike the immediate catastrophe of nuclear weapons or the relentless blow of climate change, the so-called threat of artificial intelligence remains only in the realm of science fiction. For example, HAL-9000, Skynet, and Ultron are all familiar villains, who are originally artificial intelligence, but betrayed their creators.

*On November 2, British Prime Minister Rishi Sunak attended the AI Security Summit at Bletchley Park in the United Kingdom. Both real AI skeptics and fake AI skeptics have been blamed for causing panic among the public and the government, as can be seen at the recent Bletchley Park event. *

The current state of AI is very different from the real problems that research scientists are trying to solve. The term artificial intelligence itself encompasses a large number of scientific fields, technological innovations, artifacts, and human activities. However, due to the deviation of public opinion, society is full of distortions and misuses of artificial intelligence.

In just a few years, AI models have grown rapidly, but these misleading threat narratives do not provide scientific inferences about future trends. No single technological growth curve grows indefinitely, and growth is also constrained by the laws of physics, energy constraints, and paradigm constraints, as we see in the production of genetically modified crops, the transistor density of semiconductor chips, and the FLOPS (performance) of supercomputers. There is no evidence that current software, hardware, or mathematics will propel us towards artificial general intelligence (AGI) without major paradigm disruptions in the future. Compared to the potential risks of gene editing in all organisms, the risks of Transformer-based AI, the cornerstone of AI chatbots like ChatGPT, pale.

There are fundamental holes in the reasoning and conclusions of AI apocalyptics – there is a huge chasm between hypotheses and justifications. Imagine if someone invented a bicycle that could quickly increase their speed in a short period of time through exercise and training. With an electric motor and lighter materials, the bike can go faster. If this bike can take off quickly, can you believe it?

It's not hard to see the absurdity of this reasoning, but that's exactly what the current public opinion is about AI: AI can become an encyclopedia through generative pre-trained models, Transformers. Next, AI will make the leap to AGI. It then becomes an artificial superintelligence (ASI) with emotional intelligence, awareness, and the ability to replicate itself. Then, there will be a giant leap forward for artificial intelligence – according to some of the scenarios presented in a recent debate in the Oxford University Alliance, artificial intelligence will target humans, and without deterrence, it will be able to use sci-fi methods to wipe out humans, such as making vegetation release toxic gases, or finding ways to deplete the sun's energy.

Each of these leaps requires a breakthrough in science and technology, which is likely to be difficult to achieve. Because many of the assumptions involved in these leaps are logically untenable. But the risks of these stories can spark the public's imagination.

As we saw recently at the AI Safety Summit in Bletchley Park, UK, these AI skeptics – intentionally or unintentionally – are ignoring the obligation to make scientific arguments, causing panic among the public and governments. The regulations that are currently being pushed are not designed to prevent ridiculous existential risks, but rather to undermine the open-source AI community that threatens Big Tech's profits. Over-regulation to raise the cost of AI development will only benefit a few wealthy people.

Ironically, the "existential threat" theory also ignores the role of human beings. Behind disasters such as Chernobyl and the tragedy of the explosion of the space shuttle Challenger is not technology, but human management systems. In contrast to the physical sciences, which deal with the real world, the field of artificial intelligence is primarily digital. Compared to physics, chemistry, biology and any other technology that directly experiments with the physical world, humans need to be more involved in the process of interacting with artificial intelligence, and find more opportunities to examine and control artificial intelligence.

The rhetoric of AI doomsday theorists obscures the fundamental, superior benefits that scientific progress and technological revolution bring to society and civilization, so much so that it can barely inspire the public to understand and use science. History is replete with examples of technology being a catalyst for human progress rather than a broomstick. Tools such as compasses, books, and computers have led human wisdom from the deep sea to the boundless universe.

The claim that AI is a threat depends on whether AI can surpass human intelligence, a concept that lacks clear measurements. Many inventions—such as microscopes and calculators—have surpassed some human capabilities, but people are excited about these inventions rather than fearing that they will exterminate humanity.

In fact, AI is ushering in a "renaissance" of the 21st century, fundamentally changing the way we acquire knowledge and solve problems. Unlike the Renaissance, which gave birth to enlightenment and advocated rational discovery of scientific truth, this era is bringing us into an age of empowerment.

The historical Renaissance benefited from printing technology and the publishing market, which allowed knowledge to spread rapidly in Europe and beyond. Early science constructed knowledge structures by "knowing how to think". Newton and Leibniz, among others, advocated and defined this rationalism. They and their contemporaries laid the foundation for a first-principles, methodical, and systematic edifice of science.

The science they created has evolved over the centuries through logical and methodical experiments to form hypotheses, reveal core ideas, and test theories. Modern artificial intelligence is reinventing this classic approach to problem-solving.

Today, the combination of massive datasets, advanced infrastructure, sophisticated algorithms, and powerful computing power heralds a new era of discovery that goes far beyond traditional logic. It promises to be a science characterized by thorough empiricism and AI-guided insight. Today's AI renaissance goes beyond the "how" to the "why". It provides the individual with not only knowledge, but also tools to solve practical problems, marking a shift in empirical methods and means. AI opens up possibilities in areas such as biology, genomics, climate science, and automation.

The current era is marked by a resurgence of empiricism, with AI's data processing capabilities automating the distillation, organization, reasoning, and hypothesis testing of knowledge, and learning insights from identified patterns. It opens the way for new approaches to scientific exploration, such as through extremely high-throughput digital content generation, complex simulation forecasting, and large-scale strategic optimization, at scale and speed far beyond what traditional first-principles-based methods and causal reasoning can handle.

This means that humanity has never had a real opportunity to tackle challenges such as climate change, cancer, and personalized medicine, which were previously impossible. This modern Renaissance fostered continuous learning and adaptation, moving societies and cultures from insisting on understanding everything before acting to exploring, understanding and ethically using it. In promoting humble access to knowledge and problem-solving, this mental model is similar to the empirical methodologies of the past.

Just as Prometheus stole the fire for humanity, AI has become a powerful tool for advancing human civilization, although it has not yet been fully mastered. We need humility, courage and freedom to master and use this tool.