Can AI systems be transparent and easy to understand? Is there a certain degree of technical transparency

Since about five years ago, everyone is talking about the Internet of Things (IoT) and how it will change everything through the interconnection of hundreds of millions of devices. We seem to go through a cycle of hype about a new technology, and now it’s artificial intelligence (AI).

In my career in the electronics industry and semiconductor for about 33 years, I have seen three major changes in the technological revolution. First is the era of microprocessors, then the Internet, and then the era of mobile. Today, as Synopsys co-CEO Aart de Geus said recently: “Now is the age of AI!” He also mentioned in a recent event that AI will promote the semiconductor industry in the next dozens of years. Years of development, because big data requires machine learning, and machine learning requires more operations, these will generate more data.

Indeed, AI is now in the ascending phase of what the market research firm Gartner calls the "hype cycle", but it seems to be different from the previous introduction of new technologies: I have never seen a technology trigger so much ethics. Arguing. AI will change many things. Autonomous vehicles, military and industrial drones, robots, and many other applications in medical, government, and urban functions may be affected.

2017 Gartner new technology maturity curve

The British government recently released a 183-page "AI in the UK: ready, willing and able?" report, covering many issues related to AI system responsibilities, supervision, and ethics. And other topics such as innovation, investment and skills in AI research and commercial applications.

Lord Clement-Jones, Chairman of the Special Committee of the House of Lords, said: “The UK has leading AI companies, a vibrant academic research culture, an active entrepreneurial ecosystem, and a large number of legal, ethical, financial and language advantages. We should make full use of it. Such environmental advantages, but the key is that ethics must be the focus of the development of AI."

Jones said: "AI is not without risks. The ethical principles proposed by the committee will help mitigate these risks. The ethical approach ensures that the public believes in the technology and sees the benefits of using it, while also preparing it to question whether the technology has been abused. . We want to make sure that this country is still an outpost for research and development of this new technology. However, it may be difficult for startups to scale up on their own."

In the committee's report, many recommendations point to the need for transparency in AI. When AI is used to make important or sensitive decisions, a spontaneous mechanism should be established to inform consumers. The report also pointed out that when the AI ​​system malfunctions or causes harm to users, it is not clear whether the current law is sufficient to clarify the relevant responsibilities. Therefore, the field urgently needs to clarify the responsibility as soon as possible.

Make AI transparent

Can AI systems be transparent and easy to understand? Is there a certain degree of technical transparency that allows people to question why the system made a particular decision?

The industry has discussed this and the accountability system of AI systems in detail. In many deep learning systems, feeding information through many different processing layers and getting the final answer or decision may make the system look like a "black box". Even its developers may not be able to determine which factors cause The system determines that one of them is more important than the other.

Timothy Lanfear, director of Nvidia's Europe, Middle East, Africa (EMEA) solution architecture and engineering team, put forward another view. He said that machine learning algorithms are usually shorter and simpler than traditional software coding, so they are easier to understand and detect in some ways. "We use systems that are too complex to absorb every day. AI is no different. It is also at a level of complexity that cannot be fully understood. However, the only thing you can do is break it down into fragments and find a way to test it. , And then check whether it is going the way you expect, and if not, take action."

The committee admitted that it is quite difficult to achieve complete technical transparency, even impossible for some AI systems, especially in some cases not even applicable. However, in some safety-critical scenarios, the technology must be transparent. Regulators in these areas must have the authority to enforce the use of more transparent AI technology.

The British AI Commission also mentioned in its report, “We believe that the deployment of any AI system that may have a potential impact on personal life is unacceptable, unless it can provide sufficient and satisfactory decisions on its own. Explanation. For example, for situations such as deep neural networks (DNN) that cannot fully explain their decisions, it may mean that the deployment of certain specific uses must be delayed until an alternative solution is found."

Stan Boland, CEO of FiveAI, a British self-driving startup company dedicated to the development of full-stack self-driving car functions, does not agree with the black box phenomenon of AI systems proposed in the report. He said: "Our system is completely modular, which is very critical for explaining and debugging the system to improve technology and services. Any system must undergo multiple independent verifications before it can be implemented."

He added that there are different ways to adopt autonomous driving around the world. “For example, there are different ways of operating in London, UK and Mountain View, USA. Each city has different lifestyles and environmental conditions. , The modes of autonomous driving systems are also different. For us, the focus is to create solutions that meet the needs of European customers."

Countries increase investment in AI

This report by the British government also provides another perspective for global AI research and development. According to data from Goldman Sachs, between the first quarter of 2012 and the second quarter of 2016, the U.S. investment in AI reached 18.2 billion U.S. dollars, while mainland China and the United Kingdom were 2.6 billion U.S. dollars and 850 million U.S. dollars, respectively. As mainland China strives to become a global leader in AI, it is expected that by 2030, investment in its AI ecosystem will be up to US$150 billion.

In view of the differences in existing resources, the scale of investment in the AI ​​field in the United Kingdom may not be as large as that of China and the United States. Germany and Canada are more comparable. Germany's AI strategy is deeply influenced by its strategy of building a flagship industry 4.0. Its strategy explores the use of AI to improve the process of smart manufacturing and produces smart products that integrate AI functions, such as refrigerators and cars. Canada’s AI strategy is less focused on the development of AI for specific areas, but the government of that country has spent US$125 million to establish three new AI research institutes and attract more AI researchers from around the world.

Processor bottleneck?

In the past ten years, although deep learning has played an important role in the progress of AI, it also has some problems. Deep learning requires a large amount of data sets, which is extremely difficult to obtain and expensive, and also requires a lot of processing power. The report mentioned that although deep learning has recently improved significantly with the increase in processing power, Moore's law has begun to face challenges, and the increase in processing power has slowed down the pace of price cuts. Innovations such as quantum computing have not yet been able to revive or accelerate it, but it is still too early, and there are still many uncertainties in the future.

Deep learning pioneer Geoff Hinton warned that the deep learning revolution might end soon. Other people's views are more optimistic, especially because of the emergence of various custom AI chips, such as Google's tensor processor (TPU), and the progress of quantum computing, which provide new development momentum for the future of deep learning.

Criminal abuse and supervision

The report also explored in depth the areas of criminal abuse, supervision and innovation. In the field of "adversarial AI" (adversarial AI), researchers effectively grasp how AI systems work, and try to make misclassifications or decisions without hiding other AI systems. In particular, the image recognition system proved to be extremely vulnerable to this type of attack. For example, researchers can cleverly change pictures or even 3D models or logos to make them look indistinguishable from the original images, but still hide them from the AI ​​system to recognize them as completely different objects.

Other examples include using false images to cause cars to collide, stop suddenly, or trigger automatic weapon launches. Of course, these situations may also occur in non-machine learning systems (as well as human decision makers), but non-machine learning methods can be used to query, recover, and debug the inferences involved, which many machine learning systems cannot do. To these.

In terms of supervision, the report pointed out that "Although the regulations on technology application are not perfect, it has promoted faster experimentation and innovation on the other hand, including the use of data and AI technology."

The degree of liability that companies and organizations face when they violate the regulations is also included. Nvidia’s Lanfear said that although the company’s employees know its ethical principles and how to comply with the regulations, he admits that it’s not easy to answer this question because “as a technical expert, this is not my core idea.” In this regard, many people agree with Lanfear. Have the same feeling. Therefore, some mechanism must be found to ensure that the current trend of ethics is not simply transformed into meaningless black box operations.

Today, the AI ​​technology we have can be used to promote the next wave of computing development based on high-end AI processors and chips. However, although the semiconductor and computing industries continue to challenge their limits, the industry will ultimately determine how much the technology can be applied in the real world-we always hear social protests only after seeing the casualties of autonomous vehicles. Discussions and disputes about AI ethical standards will help us think about how technology can be deployed safely and effectively, and ultimately can be accepted and adopted by society.

7 Inches Tablet PC

The 7-inch tablet can be used as the golden size of a tablet computer. It is small and portable. It can be used at home and outdoors. You can browse the web, watch videos and play games. It is a household artifact. Although the size of the 7-inch tablet is inclined to the tablet, the function is more inclined to the mobile phone, so it can also be used as a substitute for the mobile phone. Compared with other sized tablets, the 7-inch tablet has obvious advantages in appearance and weight. Both the body size and the body weight have reached a very reasonable amount.

1.In appearance, the 7 inch tablet computer looks like a large-screen mobile phone, or more like a separate LCD screen.

2.In terms of hardware configuration, the 7 inch tablet computer has all the hardware devices of a traditional computer, and has its own unique operating system, compatible with a variety of applications, and has a complete set of computer functions.

3.The 7 inch tablet computer is a miniaturized computer. Compared with traditional desktop computers, tablet computers are mobile and flexible. Compared with Laptops, tablets are smaller and more portable

4.The 7 inch tablet is a digital notebook with digital ink function. In daily use, you can use the tablet computer like an ordinary notebook, take notes anytime and anywhere, and leave your own notes in electronic texts and documents.

7 Inches Tablet Pc,Quad Core Tablet 7 Inch,7 Inch Gaming Tablet,Supersonic Tablet 7 Inch

Jingjiang Gisen Technology Co.,Ltd , https://www.gisengroup.com