清华周道许：AI于金融是一把“双刃剑”

Core Viewpoint - Artificial intelligence (AI) is reshaping the financial industry by enhancing efficiency and effectiveness, while also introducing challenges such as algorithmic opacity, data gaps, and new systemic risks [1][4]. Group 1: AI Governance in Finance - The governance of AI in finance should focus on being controllable, trustworthy, and sustainable, addressing issues like algorithm compliance and transparency, data governance, and risk monitoring [4][5]. - Establishing an "algorithm compliance and transparency" mechanism is crucial, requiring AI decision-making processes to be traceable and explainable, especially in areas like credit approval and risk assessment [5]. - Data governance and privacy protection are essential, necessitating strict adherence to data security laws and the exploration of privacy-preserving technologies [5]. - A dynamic risk monitoring system should be created to identify new systemic risks arising from AI, such as model homogeneity and algorithmic resonance [5]. - Ethical norms and responsibility recognition must be improved, ensuring clear accountability for AI decisions and preventing algorithmic discrimination [5]. Group 2: Future Applications of AI in Finance - AI's role in finance is evolving from auxiliary to collaborative and even autonomous decision-making, with potential applications in core decision-making areas like investment research and credit pricing [6][7]. - The shift from "process optimization" to "decision reconstruction" indicates that AI will play a more significant role in critical financial decisions [6]. - AI will facilitate deeper integration between finance and other sectors, creating "scene financial intelligent bodies" that can automate credit and risk management processes [6]. - The transition from "passive response" to "proactive foresight" in risk management will enable early identification of financial risks, providing valuable time for regulatory interventions [7]. Group 3: New Risks Associated with AI - AI introduces new risks such as model risk and algorithmic resonance, which can lead to collective misjudgments among financial institutions [9]. - Data pollution and adversarial attacks pose significant threats to AI models, necessitating the development of robust defense technologies [9]. - Ethical issues and fairness concerns arise from algorithms amplifying existing biases, requiring fairness audits and diverse evaluation mechanisms [9]. Group 4: Regulatory Approaches - Balancing innovation and safety is essential, with regulatory frameworks needing to evolve alongside AI technology [10][11]. - The implementation of regulatory sandboxes allows financial institutions to test AI products in controlled environments while regulators observe and learn [11]. - Developing intelligent regulatory platforms can enhance oversight efficiency, utilizing AI to monitor AI systems and ensure compliance [11][12]. Group 5: Talent Development in Finance - The financial education system must adapt to cultivate talent that understands finance, technology, and ethics, shifting from knowledge transmission to capability reconstruction [13]. - Course offerings should integrate practical modules on AI and ethics, preparing students for real-world applications in AI finance [13]. - Emphasizing critical thinking and innovative leadership skills is vital for developing strategic talent capable of navigating AI's complexities in finance [13].