In the rapidly evolving landscape of financial technology, a crucial focus has emerged regarding the enhancement and application of intelligent financial systems. As highlighted by recent reports, since the beginning of 2023, nearly 80% of intelligent financial application cases have revolved around generative AI technology, with half of these cases employing a collaborative model between large and small models. However, the notable advancements in technology remain sparse and far from groundbreaking. The concentration of intelligent financial applications predominantly resides in operational management, customer marketing, and risk compliance—areas traditionally highlighted in digital finance. It raises questions about how intelligent financial innovations should differentiate themselves, given that the current understanding of "smart finance" does not significantly outshine the previously discussed concepts of digital finance and financial big data.

During a period characterized by digital finance, both China and the United States have been recognized as leaders globally, whether evaluating banking sectors, interbank institutions, financial transactions, or financial regulation. In several aspects, including the digital transformation of large commercial banks, the number and ratio of financial technology patents, and digital financial inclusion capabilities, China stands shoulder to shoulder with, if not ahead of, the United States. For instance, a paper by the Federal Reserve noted that the digitization of interbank market trading infrastructure in the U.S. has not progressed at the pace of its Chinese counterpart. Data from China’s National Intellectual Property Administration reveals that the share of financial technology patent applications from China stands at a remarkable 44.3%, while the U.S. claims only 20.2%. While the sheer number of fintech patents may not fully encapsulate the broader industry landscape, they do provide a window into the development dynamics of China's financial technology sector.

The potential of intelligent finance, specifically the role of artificial intelligence in solving complex problems in the digital age, cannot be overstated. Take, for instance, the application of AI in the healthcare sector. AI technologies have the power to tackle known intricate health issues, revealing complex associations that were previously acknowledged but hard to validate in a timely manner. Scientists have long understood that the combination of just 20 amino acids can yield a variety of proteins, yet the traditional approach of generating these proteins one by one is painfully slow. With the help of AI, systems like AlphaFold2 have predicted the three-dimensional structures of proteins for over 100 million species, marking a significant advancement in the identification of biological structures. Moreover, AI can uncover unknown health challenges by enhancing drug discovery processes, which historically relied heavily on experimental screening with limited efficiency and success rates. Companies like Insilico Medicine have used AI to design new anti-aging drugs in just 46 days, demonstrating a breathtaking acceleration in pharmaceutical innovation. Lastly, AI enables the analysis of diverse and granular data sets to provide personalized medical services, a stark departure from traditional models that often fail to account for the complexities of individual patient needs.

The capability to analyze extensive multimodal real-time data is paramount in advancing financial research and addressing complex financial challenges. The dynamics of market behavior—shaped by investors, consumers, and businesses—constitute a complex system that was difficult to encapsulate with previous limited data sets. The explosion of data in the digital age allows for unprecedented volumes and dimensions of behavioral analysis. Financial decision-makers are often not fully rational and are susceptible to influences such as emotions, situational context, psychological biases, and unforeseen variables. Measuring these complex factors was challenging in the past, but today's high-frequency data, like those extracted from media platforms, contain valuable insights into participant emotions and sentiments. Recent Chinese economic policies have highlighted "confidence and expectations," a quintessentially complex issue that requires nuanced understanding and analysis. Developing more effective AI models that can uniformly process multimodal data—including textual information—is crucial for enhancing the anticipation and decision-making related to monetary policies.

Moving forward, it is critical for the financial sector to accelerate the exploration of advanced pre-trained deep learning models in financial analytics. The urgency of this endeavor stems from the scale effects of high-level models where early leaders dominate increasingly, creating a widening gap for those lagging behind. Traditional economies of scale are giving way to increasing returns in the era of data and AI, where the surge in model scales leads to enhanced capabilities that outstrip those available to smaller models. As model sizes grow, their performance improves significantly, leading to the emergence of novel capabilities that are impossible for smaller models to attain. The recent OpenAI conference underscored this by demonstrating reinforced fine-tuning that could elevate models to expert levels in specialized fields with minimal data, marking a decisive step away from merely memorizing answers toward fostering reasoning within specified domains.

Pioneering figures in this field are making rapid strides toward breakthroughs. AI has surpassed human performance across numerous benchmarks, including image classification and language understanding. However, in more sophisticated tasks like competitive-level mathematics and complex reasoning, AI still trails behind human counterparts, even as its learning speed accelerates. Sam Altman's concept of the "law of accelerating returns" suggests that the quantity of intelligence in the universe doubles approximately every 18 months, capturing the essence of AI's rapid evolution and its transformative implications for socio-economic structures. Looking ahead, as decision-making challenges grow increasingly intricate, the integration of large models will become ever more significant. The advancements in intelligent financial technologies must strive beyond mere incremental improvements and aim to reshape the landscape fundamentally.

A critical point of further investigation concerns the release and reuse of financial data. The lifeblood of intelligent systems is data, yet the reports so far have not delved deeply into this subject, possibly presuming that financial institutions can develop intelligent applications using their proprietary datasets. While this assumption holds in a digitized context, the demand for data during the intelligent transition is exponentially increasing. Without access to vast datasets for constant reuse, the intelligence development process can face significant impediments. With China's extensive reservoir of high-quality financial data, harnessing and optimizing this information could catapult it to the forefront of global intelligent finance.

The relationship between data challenges and the advancement of large models and other frontier technologies is crucial. The global competition in the data-driven industry necessitates maximizing the utility of resources to enhance competitive advantages in the intelligent era. Despite China's lead in digital finance, the current lag in intelligent finance can be largely attributed to a lack of robust open-source datasets—a critical constraint on developing large models. The training data utilized by major model providers can encompass open-source datasets, web crawled data, commercially acquired and licensed data, proprietary business data, and synthetic data. Preliminary reports indicate that top Chinese firms possess a notably higher share of proprietary data, while the accessibility of open-source datasets and web-sourced data remains limited and costlier in comparison. Furthermore, a greater reliance on commercial partnerships for data acquisition affects the efficiency of model training.

Access to high-quality, voluminous data is a fundamental prerequisite for the advancement of intelligent finance. Ensuring safe, anonymized data sharing and supply is paramount. For instance, one company has developed an extensive data collection and annotation platform to facilitate comprehensive financial asset allocation services, marking a significant effort in meeting model data requirements. The burgeoning demand for extensive data in cutting-edge models underscores the importance of ongoing data unlocking and reuse. Modern financial institutions are committed to enhancing their 'data augmentation businesses,' potentially monetizing data access for a broader range of sectors to foster innovation under the premise that "data empowers AI." While financial datasets are often sensitive, numerous technological safeguards ensure data security, allowing for the realization of significant concepts such as "data never leaves the domain, visibility is optional." Maintaining data security while fostering a culture of collaboration and interactivity is crucial.

Moreover, it is essential to balance commercial interests with national objectives to guide the trajectory of intelligent finance. China's large state-owned commercial banks bear responsibility not only for their commercial imperatives but also for broader monetary policy goals. As such, directing intelligent finance to prioritize efficiency and risk management is vital. Innovations in products, business processes, and customer service should ultimately serve to bolster efficiency and control risk. Simultaneously, the concept of financial inclusivity, which promotes equitable distribution of financial resources, must remain at the forefront of innovation. Current intelligent finance models often derive from immense datasets that may inadvertently exacerbate resource allocation disparities, further entrenching challenges for smaller enterprises grappling to access financing. Addressing information concentration issues is imperative for rectifying the balance between stronger and weaker entities.

Furthermore, attention must be placed on the new risks and imbalances introduced by intelligent technologies. Innovations in intelligent finance inherently harbor technological and operational risks that could potentially trigger systemic vulnerabilities. Notably, the growing homogeneity among quantitative models employed by financial institutions can lead to synchronized strategies that heighten risks. Traditional financial markets' diversity in participant expectations and risk appetites contributed to effective risk dispersion. In contrast, the current reliance on uniform data sources and aligned quantitative models can facilitate the convergence of asset strategies, which in turn, amplifies the risk landscape. The explosion of digital social networks and content-driven platforms has intensified the influence of narratives on investor sentiments and market emotions, particularly amplifying congruent expectations among individual investors. These elements collectively threaten to escalate volatility within financial markets, highlighting the complexity and interconnectedness of the modern financial ecosystem.