On November 8th, the Wuzhen Summit of the World Internet Conference opened. On the opening day, Bu Xianlu, Vice President of Cisco Greater China and General Manager of the Security Division, received an exclusive interview with reporters from Nandu. This is the third time that Bu Xianlu has participated in the World Internet Conference Wuzhen Summit. He told reporters from Nandu that his impressive topics during the epidemic were "resilient supply chain" and hybrid office. This year, when he came to the Wuzhen Summit, he found that everyone's focus was on the development of innovative technologies such as AI.
Bu Xianlu pointed out that AIGC will bring both opportunities and challenges. In terms of network security, AIGC has also lowered the threshold for network attacks, and enterprises should have corresponding regulations and control measures in the process of using AIGC.
Bu Xianlu, Vice President of Cisco Greater China and General Manager of Security Business Unit
AI significantly reduces the threshold and difficulty of network attacks
Nandu: How many times have you been to the Wuzhen Summit? How is it different from what I used to feel?
Bu Xianlu: This is my third time attending the conference. In the past, I have been deeply impressed by topics such as supply chain resilience, business resilience, hybrid office, etc. This year, I have talked more about innovation and development. In the past two years, with the breakthrough of AI technology, there has been more discussion on this aspect.
I also watched a lot of innovative achievements released today. The innovation of AI continues to make new breakthroughs in various industries and application scenarios, and so does Cisco. We have made many fundamental breakthroughs in network security and network communication applications. As a global enterprise, we invest over $6 billion in research and development expenses annually. The same is true in the field of network security, which I am familiar with and good at.
But at the same time, there are also many concerns, including privacy protection, intellectual property protection, and so on. Some experts even mentioned whether this generative artificial intelligence will bring challenges to the development of the entire human society in the future.
Cisco has conducted a lot of research on network security to ensure a responsible framework and model for AI applications in various industries. Ensure that AI itself is safe and reliable during the application process.
Nandu: What specific network security issues may AI bring?
Bu Xianlu: The biggest feature of AIGC is its new creations and inventions, including text, voice, and even video. So now we will see many fake sounds and videos, and the technical threshold is getting lower and lower. The same is true in the cybersecurity industry. In the past, when hackers or the attacking party attempted to launch an attack, the technical threshold and cost were relatively high. With the help of AI now, the threshold and difficulty have been greatly reduced.
Writing a phishing email has become more difficult to identify than before. Or when someone needs to scan and evaluate the risks of the entire enterprise, they can even automatically call it out through AI, which is simpler than before. To put it more plainly, it used to be something that could only be done by a professional hacker. Today, a computer novice can do anything he wants.
Nandu: How should companies respond to this risk?
Bu Xianlu: We have recently released a customer privacy survey report on AI, which also provides several suggestions. Firstly, it is necessary to provide training and education to employees to make them aware that AIGC not only brings many opportunities, but also brings many risks and challenges. All employees should have this awareness.
The second is to have corresponding regulations and control measures in the process of using AIGC to prevent the leakage of personal data, intellectual property, and source code within the enterprise. Because when feeding this AIGC for training, the information may have been shared.
In addition, investment should be increased from the perspective of network security. We will soon launch many new solutions in AI security to help small and medium-sized enterprises and large enterprises better cope with these AI based attacks.
Nandu: Besides bringing risks, what role can AI play in network security?
Bu Xianlu: Cisco's products include the following aspects. The first is to help enterprises automatically generate security policies and reduce the complexity of operations and maintenance; The second is to launch AI intelligent assistants in the security management center to help management personnel in the operation and maintenance center shorten detection time; Thirdly, about 80% of the current online traffic is encrypted. How to detect attacks and malicious traffic in the network without decryption is also a specific application scenario of AI. What I have provided are just a few simple examples. We have been investing in predictive AI and generative AI for many years, so we are constantly releasing new innovative results.
Cisco invested in large model enterprises six years ago
Nandu: What is Cisco's layout on the big model?
Bu Xianlu: When we talk about AI, we will emphasize four aspects of capabilities, namely data, computing power, algorithms, and network security. From the data perspective, Cisco is a leading global network supplier, with over half of the data traffic passing through Cisco's network, so our data volume is large; In terms of algorithms, we accelerate our capabilities through self research, innovation, and mergers and acquisitions, including computational power; In terms of network security, Cisco has also invested heavily in this area. We have a team of approximately 500 scientists who specialize in researching attack techniques and vulnerabilities, and apply the research results to all of our products and services. At the same time, we open source some of the results to the community and developers. These four aspects together constitute the four core elements of AI.
In addition, at the beginning of this year, Cisco acquired Armorblox, a leading company in large language modeling. This acquisition has been completed, and we have also applied the corresponding technology to many solutions. Cisco has actually had a long accumulation of big language models, and we didn't pay attention to them just because they have been popular in the past year or two. Armorblox was actually invested by us six years ago, and we officially acquired it six years later.
Nandu: Is there any experience at home and abroad that can be used for reference in terms of compliance with large models?
Bu Xianlu: Let me share Cisco's own experience first. We first proposed the "responsible AI model" and transformed it into practical control measures to be applied in Cisco's product technology services and product development processes. This mainly includes several aspects, such as security, privacy, reliability, fairness, transparency, etc.
In different markets and countries, various industries have also introduced many AI regulatory measures. I believe that with the deepening of everyone's understanding of generative AI, the continuous breakthroughs in various application scenarios, and the practice in this area, laws, regulations, and best practices will become more and more abundant, providing protection for the innovation of the entire implementation of generative AI and ensuring that it is in the right direction.
Nandu: Do you think AI will make humanity more inclusive or widen the gap?
Bu Xianlu: I am an optimist myself. In recent years, generative AI has made some breakthrough progress, and we have also seen various industries benefit from it. Many new scenarios have emerged. But at the same time, we also need to pay special attention to the risks brought by the emergence of new technologies. In addition to security, privacy, and compliance risks, we also need to pay attention to whether it will bring about a greater digital divide.
Our Cisco Academy's plan has also won awards at this conference, and has trained over 1 million people in China. We aim to cultivate college students, engineers, and others to master leading network and artificial intelligence technologies. This is one example where such a platform can also help narrow the digital divide in the AI era.
Nandu: What opportunities and trends have Cisco observed on the path of new industrialization?
Bu Xianlu: The basic support in the new industrialization, we call it super connection. The latest technology we are talking about now cannot be separated from super connected at the bottom. There will be more and more businesses, applications, and data carried on the super connection. It is important to ensure that it has a secure and reliable network for hosting.
This presentation has become a hybrid and multi cloud architecture today, which we often hear in digital transformation. Superconnection has a hybrid multi cloud architecture, and the current form of hybrid office still exists. Therefore, we summarize it into three categories: H: type connected, Hybrid Multi cloud, and Hybrid work. Cisco can fully leverage our advantages in network and security to help businesses transform digital and provide a more secure, reliable, stable, and efficient network.