Recently, an Apple research report has sparked heated discussions, which explores the current status and limitations of generative artificial intelligence (AI) in the field of financial advice. While more and more American consumers, especially young people, are starting to use tools such as ChatGPT to seek financial advice, AI has significant flaws in complex mathematical and logical reasoning, which makes it in providing accurate finance. There are huge challenges in the recommendations. This article will analyze the report in depth and explore the future prospects of the application of generative AI in the financial field.
Recently, a research report released by Apple has sparked discussions on the effectiveness of generative artificial intelligence (AI) in financial advice. The survey shows that more and more American consumers use generative AI tools, such as ChatGPT, to obtain financial advice, a trend that is particularly evident among young people. According to a Motley Fool survey, 54% of Americans have sought recommendations for financial products through ChatGPT, with younger generations being more employed.
Image source notes: The image is generated by AI, and the image authorized service provider Midjourney
The survey results show that although half of consumers expressed their willingness to use ChatGPT to obtain recommendations, the proportion of interest in specific financial products is relatively low. For example, only 25% of respondents want ChatGPT to recommend a credit card for it. In addition, the respondents' recommendations for ChatGPT were generally "somewhat satisfied" and the average satisfaction was 3.7 in the five-point scale, showing some recognition.
However, Apple's research points to the significant flaw in current large language models (LLMs) in logical reasoning, especially mathematical reasoning. Researchers found that these models perform poorly when facing complex mathematical problems and often fail to correctly understand or solve simple mathematical calculations. As the complexity of the problem increases, the performance of the model further declines, showing deep-seated problems in its inference process.
An article from TechCrunch lists multiple examples of generative AI errors in mathematical calculations, explaining their shortcomings in dealing with basic mathematical problems. The report mentioned that the "blocking" technology used by AI models when processing numbers often destroys the relationship between numbers, resulting in calculation errors.
In addition, machine learning also faces challenges when dealing with financial advice. Although some people confuse machine learning with statistical analyses such as regression analysis, machine learning actually requires a decision-making process, an error evaluation function, and a model optimization process. This makes it possible for generative AI to effectively meet user needs in financial advice.
Apple's research shows that banks and credit unions should not rely on AI for financial advice at the current stage. Although there may be some improvement in the future, generative AI will still be difficult to compete with complex financial consulting for the foreseeable future.
Key points:
54% of Americans have obtained financial advice through ChatGPT, and the younger generation is more likely to use.
Apple research shows that generative AI has significant flaws in mathematical reasoning, especially incorrect handling of complex problems.
At present, banks and credit unions should not rely on AI to provide financial advice, and it may take 5 to 10 years to improve in the future.
To sum up, although generative AI has shown certain potential in the field of finance, its shortcomings in logical reasoning and mathematical calculations limit its application in complex financial consulting. In the future, AI technology needs to overcome these limitations to truly be competent for the work in this field. At present, it is wise to use it with caution.