Artificial intelligence opens the era of great change, and behind the opportunity is vigilant against the "undercurrent"

21st Century Business Herald reporter Wu Bin reported from Shanghai

Artificial intelligence is on the rise.

On July 17, local time, James Rensler, chairman of the US Securities and Exchange Commission (SEC), delivered a speech entitled "From Newton to Artificial Intelligence", emphasizing that artificial intelligence is the most transformative technology of this era, which can be compared with the Internet and automobile mass production.

When the plague swept through Europe and England in the 1660s, Newton pondered esoteric things like gravity and calculus. Shortly after the latest outbreak, OpenAI released GPT-3 in June 2020, GPT-4 on March 14, 2023, and just two days later, on March 16, Baidu also launched ERNIE Bot. In Jensler's view, both artificial intelligence and Newton's work on gravity are based on data and computing power.

From the "father of artificial intelligence" Alan Turing's 1950 paper "Computers and Intelligence", to IBM Deep Blue and Google AlphaGo defeating the strongest human chess players, artificial intelligence has entered all aspects of life: including natural language processing, translation software, recommendation systems, radiology, call centers in the financial system, compliance procedures, trading algorithms and other fields.

As the world embraces the new era of artificial intelligence, what opportunities and challenges will there be in the future? What are the megatrends in the development of artificial intelligence?

Semiconductors will become a trillion-dollar industry

Semiconductors play a crucial role in the development of artificial intelligence, and AI chip giant Nvidia has soared more than 200% this year, becoming the big winner of the artificial intelligence boom.

Anjali Bastianpillai, senior manager of thematic equity client portfolio at Pictet Asset Management, told the 21st Century Business Herald that AI is a particularly resource-intensive technology that requires massive data and processing power to create new content. Semiconductors are rightfully the main beneficiaries of AI, and large language models also require sufficient memory capacity and bandwidth, and memory, computing and storage semiconductors account for the majority of semiconductor sales.

Historically, semiconductor companies have benefited from increasing industry consolidation, driven by economies of scale and resource concentration in response to rising costs of resource-intensive R&D, design, and manufacturing; 5G, the huge demand for the Internet of Things; ensuring supply chain security; expanding the product portfolio beyond microchips, such as software products; Growing demand for custom chips and foundries; Access to highly specialized talent; The need for cloud and data center management technologies.

In Anjiali's view, semiconductors are the backbone of technological innovation, powering everything from smartphones and laptops to servers and supercomputers, and semiconductors are expected to become a trillion-dollar industry by 2030.

Under the AI boom, there is now a shortage of GPUs around the world, and the market expects this shortage to continue at least until next year.

Angeli believes that AI has increased the need for faster and more efficient computing, driving the development of cutting-edge technologies and related production and design capabilities, while creating a higher barrier to entry. At the same time, semiconductor equipment manufacturers and software companies that support semiconductor manufacturing will also benefit from the rise of AI.

Embodied intelligence is the trend of the times

This year, general artificial intelligence represented by ChatGPT has achieved a huge breakthrough, and its powerful language understanding and dialogue capabilities are astounding.

After all the shocks of generative AI, the industry generally believes that the next big challenge is to achieve embodied general artificial intelligence. In the future, general artificial intelligence should also have flexible "limbs" and "bodies", and have keen and flexible senses in order to have the ability to interact with the world at a deeper level.

NVIDIA founder Jensen Huang said at the ITF World 2023 semiconductor conference that embodied intelligence is an intelligent system that can understand, reason and interact with the physical world, which is the next wave of artificial intelligence.

Xu Huazhe, assistant professor at the Institute of Interdisciplinary Information Studies at Tsinghua University, told the 21st Century Business Herald that embodied intelligence can be regarded as the ultimate pursuit of strong artificial intelligence or general artificial intelligence, which is very close to the robots written by science fiction writers such as Asimov, which should be able to complete hundreds of tasks in a large number of different scenarios. The realization of embodied intelligence can allow people to have a robot friend and serve people, such as serving tea and water for people in the home scene, washing and cooking, and completing a series of service tasks. However, a key point here is that everyone's home layout is different, and these differences, which humans may usually ignore, are a huge challenge for embodied intelligence.

Compared with general artificial intelligence represented by ChatGPT, embodied general artificial intelligence faces many new challenges. Xu Huazhe analyzed that in terms of AI, data acquisition is a difficult point, requiring rich multimodal data, which is much more difficult to obtain than text. In terms of training methods, the behavior must conform to the laws of the physical world, and the plain text pre-training mode is no longer applicable. In terms of robots, the hardware carrier of embodied general artificial intelligence needs to be a general-purpose robot such as a humanoid robot, which is significantly more difficult to develop and manufacture than ordinary robots. General-purpose robots need to integrate the skills required by all robots, such as walking, navigation, and operation, and work in unknown environments.

Behind the opportunity, be wary of "undercurrents"

On the whole, the current artificial intelligence is still in its infancy, and there is still huge room for development in the future, at the same time, reducing the negative impact of artificial intelligence has become an important issue, and the world needs to "prevent problems before they occur".

Emad Mostaque, CEO of Stability AI, noted that AI technology is still in the early stages of development, and "AI is not quite ready for large-scale deployment in large industries such as financial services, but we can see its value." ”

There is also an "undercurrent" behind the opportunity. Jansler warned that the next financial crisis could come in the use of AI by companies, and he warned of the potential "systemic risk" posed by the spread of AI technology.

Currently, some financial institutions are already using AI in a variety of functions, including tasks such as assessing new customers or checking suspicious transactions, but Jensler cautioned that despite the potential to improve efficiency, these systems should be scrutinized because of the potential for biased decisions. Financial advisors and brokerages may also use AI to guide customers to buy products, which is a conflict of interest, and the SEC plans to crack down on these issues by considering new rules as early as this fall.

It should be noted that the US Federal Trade Commission (FTC) has launched an investigation into OpenAI, questioning whether its popular artificial intelligence conversational robot, ChatGPT, poses a risk to consumers' reputation and data, which is the first official investigation of the technology.

Companies like ChatGPT will make every aspect of people's lives smarter, but the potential for "runaway" is just as dangerous. Federal Trade Commission President Lina Khan, who testified before Congress on July 13, previously raised concerns about AI, saying that law enforcers "should be vigilant early" about transformative tools like AI.

In addition, in June, the EU agreed to introduce a draft AI law that would require companies such as OpenAI to disclose AI-generated content, help distinguish so-called deepfake images from real images, and ensure protection against illegal content.

Looking forward to the future, mankind will one day make greater breakthroughs in embodied general artificial intelligence, how do you view the opportunities and risks?

In Xu Huazhe's view, the development of embodied artificial intelligence will have a fundamental change in the operating mechanism of the entire society. The first is the positive side, with the aging and labor shortage, the pension problem has become a hot issue in society. Embodied intelligence can effectively provide high-quality labor, and take care of the elderly, monitor diseases, and serve disabled people in home scenarios. On the other hand, humans with their hands free will have more time to enjoy life or develop creative work and tasks.

But embodied intelligence also brings new societal challenges. Since robots may come into direct contact with people, their safety needs to be strictly guaranteed. On the other hand, it is also necessary to ensure that the algorithm does not do evil, if the algorithm itself is full of bias and malice, there will be a risk of irreparable trauma to the entire human society. Therefore, robots need to accurately understand human intent and work in tandem with humans, in line with human values.

For more information, please download 21 Finance APP