This article originally appeared in Digital Wealth News
Artificial Intelligence (AI) has transformed numerous facets of our lives, including how we access and use financial services. Among the many applications of AI in the wealth industry, image generation stands out as an area that promises significant potential but also raises questions about trustworthiness. Can AI be trusted to generate images for the wealth industry? The Digital Wealth News team met with Teresa Leno, CEO and Founder of Fresh Finance, to explore this complex topic.
The advent of AI capabilities such as Generative Adversarial Networks (GANs) has created highly realistic images. These can be used to visualize data, market trends, asset portfolios, digital marketing materials, and a whole range of other applications in the wealth sector. AI’s ability to quickly, accurately, and efficiently generate such images is valuable.
AI is already being used to collect client performance data from a data set within a wealth management system and then create bar graphs and pie charts- a capability that exists today. Visually illustrating data benefits the client and advisor because it’s appealing and easy to understand.
“I believe Fintech providers have a BIG opportunity to advance AI-generated imagery even more within their solutions for client reporting purposes.” – Teresa Leno, CEO and Founder of Fresh Finance.
However, alongside these advantages of AI imagery come emerging concerns about reliability and trust.
AI ‘tool training’ is critical.
“AI-generated images are subject to the data on which they are trained. The output may have flaws if the dataset contains bias, misinformation, or errors. In the wealth industry, where decisions about wealth management, investment strategies, and financial planning are often made based on graphs, charts, and other visual information, any error or misrepresentation may lead to significant losses. Therefore, verifying the data for training AI systems becomes crucial,” says Leno.
The wealth industry deals with sensitive financial data, and maintaining confidentiality and privacy while using AI to generate images is imperative. Safeguards must be in place to ensure that the AI does not inadvertently reveal confidential data in its output. This could involve obscuring specific details in the generated image or controlling the data the AI can access.
Leno provides a more straightforward example of using AI to create an image accompanying a blog post about saving for one’s retirement nest egg. Without training, the AI tool will likely make an image such as a golden egg or a U.S. hundred-dollar bill—neither of which can be represented in any communication with the public.
“Another aspect to consider is AI’s accountability. Unlike a human designer who can correct any inaccuracies or mistakes in the images they create, assigning responsibility when AI generates inaccurate or misleading visuals is more challenging. This poses a serious concern for the wealth industry,” adds Leno.
While these potential pitfalls may seem daunting, they do not necessarily mean that AI cannot be trusted to generate images for the wealth industry. Instead, they highlight the importance of diligent compliance oversight and risk management strategies. Implementing strict data verification procedures, investing in high-quality training data, applying rigorous privacy settings, and establishing clear protocols for human review and approval of AI-generated visuals must occur.
Leno comments that AI image-generation tools are behind ChatGPT tools from a compliance perspective. She goes on to say that, as of now, she’s not tested any AI-image generation tools she would want to integrate into Fresh Finance’s marketing communication software.
In conclusion, AI’s trustworthiness in generating images for the wealth industry hinges on many factors, including data quality, privacy preservation, and accountability assignment. While challenges exist, with management and ongoing technological advancement, AI has the potential to play a pivotal role in improved creation and accuracy of imagery.