Reporter Liu Hui.
The Nasdaq 100 index, which tracks technology stocks, has around 47 so far in 20235%。In addition to Google, OPE?, the global artificial intelligence companies that have attracted attentionNAI, Apple, Microsoft, Amazon, IBM, Facebook, Tesla, NVIDIA, Deepmind, etc., Chinese AI companies include, Tencent, Alibaba, JD.com, etc.
According to data from Pitchbook, a financial data analytics company in the venture capital market, the amount of investment in AI-related startups reached about $68.7 billion in 2023. But after experiencing the "huge earthquake" of OpenAI's personnel, the industry is also thinking about how technology companies can balance public interest and business successHow to improve the corporate governance of technology companies?
Make sure the AI moves in the direction of good intentions.
Stéphane Sireau, global vice president of high-tech industry at Dassault Systèmes, said in an interview with China Economic Times that in the past ten years, technology has been accelerating, especially AI technology has accelerated the change of the world, but we must be very careful to deal with it, and make AI technology more reliable and safer to be applied to industrial platforms, so that AI technology can bring value to enterprises.
Li Wei'an, chair professor at Nankai University, dean of the China Institute of Corporate Governance, and vice chairman of the expert advisory committee of the China Association of Listed Companies, told the China Economic Times that AI technology is a double-edged sword, which brings risks and challenges while changing production methods. In particular, the emergence of ChatGPT marks a new milestone in artificial intelligence technology, providing intelligent and personalized service solutions for humans, but it also raises concerns about the potential risks and ethical issues of artificial intelligence technology. The moral choice of whether the decision-maker's starting point is to achieve a beautiful vision of safety and benefit for all mankind or to commercialize it determines whether AI will develop "for good" or "for profit", and this moral choice will profoundly shape the evolution of AI, not only affecting the use of technology, but also directly related to the fate of human civilization. "We must carefully guide the development of AI, considering multiple factors such as data privacy, fairness, security, etc., to ensure that we move in the direction of good intentions. ”
At present, the promotion of AI governance and AI corporate governance has become a global trend. On November 1, 2023, the first Global AI Safety Summit opened at Bletchley Estate in the United Kingdom, and the first global AI statement jointly reached by participating countries, including China, was officially released, agreeing to establish an AI regulatory approach through international cooperation.
On October 30, 2023, the White House released the latest executive order signed by Biden, the Executive Order on Secure, Secure, and Trustworthy AI, to ensure that the United States is leading the way in grasping the promise of AI and managing its risks. The executive order includes eight goals, including establishing new standards for AI security, protecting the privacy of Americans, and promoting fairness and civil rights.
Li Wei'an believes that the establishment of an internal and external governance system for AI enterprises should be accelerated.
First of all, we need to promote the construction of AI enterprise board of directors, and matters related to enterprise AI should be included in the board of directors for decision-making and become a decision-making direction of enterprises. Establish a special ethics committee in the board of directors, and clarify the composition and operation of the AI ethics committee in the articles of association, and judge and grasp the direction of technology application. Establish a Chief AI Officer or Chief Ethics Officer to spearhead the development of a dedicated AI development strategy.
Second, establish an internal data compliance review mechanism and internal control mechanism to ensure security and compliance in R&D and data use. Set up a compliance department and establish a risk assessment and prevention mechanism for data security issues.
Thirdly, a comprehensive talent training program will be implemented to enable management personnel and R&D personnel to understand the ethics and social responsibilities of AI technology, enhance their understanding of the company's values, and cultivate a corporate culture that can strike a balance between technology and ethics.
Finally, we need to comprehensively upgrade the governance system and strengthen green governance and emergency governance. AI corporate governance should break through the boundaries of traditional corporate governance that simply pursues the maximization of economic benefits, and should assume more responsibility for enterprises as a platform for social resource allocation.
will be responsible for establishing the socialization goal of promoting the sustainable development of society.
Why Ant HR Opic designed the governance structure of the "Long-Term Interest Trust".
Employees who left OpenAI founded another AI unicorn, Anthr Opic, in 2021, likewise pledging in its articles of association to "prioritize helping humanity while maximizing profits." However, Ant HR Opic said that there is no contradiction between the public interest and commercial success or shareholder returns, and that there is often a strong synergy between the two. To this end, Anthr Opic has devised a governance structure called the "Long-Term Interest Trust": the company is overseen by a board of directors. The Board selects and oversees the leadership team, specifically the CEO, which is responsible for hiring and managing employees. Shareholders have the right to elect, remove and sue directors, while board members do not. In order to align the incentives of directors with the interests of shareholders, directors' remuneration is usually paid in the form of **. In order to monitor the decisions of the board of directors in the interests of all mankind, Anthr Opic has set up a trust to supervise the directors. The Trust is made up of five independent trustees who have no financial interest and have the power to determine a majority of seats on the Board of Directors.
Li Wei'an said that as a start-up company in the field of artificial intelligence, the company structure was relatively streamlined at the time of its initial establishment, but with the continuous development and growth of the company, the shortcomings of the governance structure and mechanism will be exposed, so it is necessary to improve the corporate governance system. Anthr Opic also aims at AI safety, but unlike OpenAI, Anthr Opic juxtaposes the public interest with the goal of commercialization rather than opposing it, expecting synergistic development within the organization, rather than achieving differentiated governance goals through the establishment of a separate subsidiary. Based on this, Anthr Opic has devised a series of checks and balances.
In terms of the governance mechanism of the board of directors, Anthr Opic's articles of association stipulate that the company is supervised by the board of directors, and the board selects and supervises the management of management, especially the CEO, who is responsible for hiring and managing employees, Li said. However, the board of directors does not have supreme power, and in order to align the interests of the board with the shareholders, Anthr OPIC stipulates that the directors are elected by the shareholders and can be removed by the shareholders. Directors are usually remunerated in the form of a company** to ensure that the interests of directors and shareholders are aligned. Most importantly, the right to elect, remove and sue directors rests solely with the shareholders, so that the shareholders' council acts as a check and balance to the board.
From the perspective of corporate governance mechanism, the board of directors represents the interests of shareholders, but whether it will be responsible to the public and other stakeholders has become a matter of great concern. Li Wei'an said.
In his view, Anthr Opic allows the board of directors to balance the public interest with the goal of maximizing shareholder value by setting up a trust. The LIC is made up of five trustees with backgrounds and expertise in AI security, public policy and social enterprise. The initial trustees are elected by the board of directors, and future trustees will be elected by the trustees by vote. This ensures that the body is sufficiently independent to balance the interests of stakeholders such as the public.
In addition, Anthr Opic created a new category (Category T) that is exclusively held by the trust at the end of its Series C funding round. Category T** gives trust companies the power to elect and remove some members of the board of directors of Anthr Opic. At the same time, Anthr Opic has also created a new board seat that will be elected by the Series C funding round and subsequent investors to ensure that there is an investor voice on the board, which is completely different from OpenAI's design. Anthr Opic has also designed an improvement process for such a governance mechanism, which has been gradually improved in practice.
Li Wei'an said that it can be seen that Anthr Opic supervises and balances the decision-making behavior of the board of directors through the trust and the shareholders' meeting. Such a mechanism is designed to ensure that the board of directors pursues the interests of shareholders while balancing safety and the public interest. Under such a governance structure, the improvement of the rules of procedure, decision-making process and other mechanisms of the governance body is particularly important, which is the guarantee of whether the governance can be effective. At the same time, the increase of governance subjects and the extension of the information communication chain will also become a hidden danger of governance, and it is necessary to strengthen the construction of information communication mechanisms and channels to ensure that trust institutions and shareholders understand the actual situation within the company in a timely manner and reduce the occurrence of problems.