The road to becoming an AI creator | Article introduction: Shocking fact! Retail is introducing #generative AI twice as fast as the financial industry! ? A thorough explanation of the reasons and the future! Check it out now! #generativeAI #retailDX #financialDX
Video explanation
How will shopping and asset management change in the future? Generative AI will revolutionize the retail and financial industries!
Hello, this is John, a familiar name for AI technology commentary! Recently, the term "generative AI" has been appearing frequently in the news and on the Internet. Many people may feel that it sounds difficult, but it is actually a very exciting technology that has the potential to greatly change our lives. In particular, the use of generative AI is rapidly progressing in the "retail industry" where we do our daily shopping, and the "financial industry" where we manage our money. However, it seems that these two industries have slightly different approaches to generative AI. Today, I will explain in an easy-to-understand manner for beginners how generative AI is used in retail and finance, what kind of future it is trying to bring, and the differences between the two approaches!
The Basics: What is Generative AI and What's its Role in Retail and Finance?
First, let's briefly review what "generative AI" is. This is AI that can "generate" new text, images, music, and even program code, just like humans. It learns large amounts of data, grasps its patterns and characteristics, and creates original content based on them. For example, it is the power of generative AI that allows chatbots to have natural conversations and automatically generate blog articles from keywords.
So what problems does this generative AI solve in retail and finance, and what are its unique features?
- Retail roles:
- Ultimate Personalization: Based on each customer's preferences and purchasing history, we will suggest recommended products and issue special coupons, providing an experience that feels like having a dedicated salesperson with you.
- Automate and improve customer support: Our 24/365 chatbots answer queries quickly and accurately, allowing human agents to focus on more complex issues.
- Automatically generate compelling product descriptions: AI will create catchy slogans and descriptions for new products, conveying the appeal of the products to more customers.
- Improved demand forecast accuracy: Analyze historical data and trends to predict future best-sellers, optimize inventory management, and reduce waste.
- Roles in the finance industry:
- Personalized Financial Advice: They propose optimal investment strategies and insurance products tailored to each individual's asset situation and life plan (however, this is being done cautiously due to strict regulations).
- Market trend analysis and forecast: We analyze vast amounts of market data to predict future price fluctuations and risks, helping you make smarter investment decisions.
- Enhanced fraud detection: AI detects sophisticated fraud patterns that traditional systems tend to miss, improving security.
- Significant improvement in business efficiency: Automate routine, time-consuming tasks like report generation, customer data analysis, and compliance checks.
According to a recent survey, the retail industry is more proactive and swift in introducing generative AI into actual services, while the financial industry is more cautious and tends to take its time in the experimental phase. According to an analysis by one AI security vendor, retail companies are incorporating generative AI into their systems 2.1 times faster than financial services companies. We will take a closer look at where this difference comes from later.
Development Speed and Resource Allocation: Retail and Finance, Different Circumstances
Naturally, introducing a new technology such as generative AI requires development resources (human resources, funds, time, etc.) There are some interesting differences between the retail and financial industries in how these resources are allocated and how development is carried out.
According to the aforementioned AI vendor report, approximately 61% of generative AI-related projects in the retail industry are in active development (high program updates and developer involvement), while in the financial services industry the figure is only 22%. This suggests that while the retail industry is moving quickly to apply generative AI to actual business, the financial industry is taking more time to experiment and verify.
Why is there such a difference?
- Retail industry motivations:
- Direct bottom line impact: The personalized recommendations and efficient customer support provided by generative AI can immediately lead to increased sales and cost savings, making it easier for management to support rapid adoption.
- Short feedback loop: When you release a new function, you can get immediate feedback from your customers, which allows you to speed up the improvement cycle and refine your service to the next level.
- Increasing competition: Being the first to offer compelling customer experiences is crucial to establishing a competitive advantage.
- Financial industry:
- Strict Regulation and Compliance: The financial industry is subject to extremely strict regulations because it handles customer assets, so careful verification is essential when introducing new technology from the perspectives of safety and compliance.
- Magnitude of risk: Since an error in AI judgment could lead to significant losses for customers, it is necessary to take every possible measure to ensure the reliability of the system.
- Complexity of existing systems: Integration with large, complex financial systems that have been built over many years can also be a hurdle to adoption.
- A long-term view of innovation: Financial institutions also tend to use generative AI for longer-term, innovative initiatives, such as developing new financial products and services in the future, rather than for short-term profits. This can lead to longer development times. In fact, the average "age" of generative AI-related code repositories (program storage locations) at financial institutions is 688 days, while at retail companies it is 453 days. This suggests that the financial industry started research and development earlier, but is taking longer to put it to practical use.
In this way, the different characteristics and priorities of each industry are resulting in differences in how they approach generative AI.
The tech behind it: How is generative AI working in retail and finance?
So how exactly is generative AI solving problems in the retail and financial industries? Let's take a look at its basic mechanism and the unique ways it is used in each industry.
At the heart of generative AI are AI models such as "large-scale language models" (LLMs) and "diffusion models" that learn from vast amounts of text and image data from the internet to gain the ability to understand the nuances of human speech and generate new content based on instructions.
- Learning Phase: First, the AI reads a large amount of textbook data (for example, product descriptions, conversation logs with customers, market news articles, etc.). In the process, it statistically learns the associations between words, image features, context, etc.
- Generation phase: Once trained, the AI generates new output based on instructions from humans (called "prompts." For example, if you instruct it to "come up with an attractive copy for this new product," it will suggest several ideas based on the training data.
The retail and financial industries have built tools and systems that apply this same basic mechanism, but each has a different approach.
Technological approaches in retail:
In the retail industry, generative AI is being used in functions that directly improve customer experience. For example,Recommended enginesanalyzes a customer's past purchase history, browsing history, and even the contents of their current cart in real time, and instantly suggests products that the customer may be interested in next.Auto-response chatbotThese systems respond to customer inquiries in natural language, as if they were conversing with a human. These systems need to have direct access to customer data and process information in real time, so response speed and processing power are key.
Tools being used include Python SDK (software development kit) provided by OpenAI and LiteLLM, which makes it easy to handle various AI models, which are becoming mainstream. This is likely due to the emphasis on increasing development speed and bringing services to market faster by concentrating on specific powerful tools. The retail industry's strategy is to "quickly develop features that have a big impact on customers using a small number of carefully selected tools."
Technical approaches in the financial industry:
On the other hand, the financial industry is more cautious about using generative AI, focusing on systems that directly impact customers rather than systems that directly impact customers.Improving internal business efficiency,Advanced data analysisFor example, they are being used as an auxiliary tool for financial analysts to prepare market reports by collecting and summarizing relevant information, and as risk management systems to detect signs of fraud from huge amounts of past transaction data.
In the financial industry, there is a tendency to try out a more diverse range of tools, such as OpenAI Client, LangChain (a framework for building complex applications by combining various AI models and data sources), and LiteLLM. This is because they value flexibility in handling a wide range of models and data formats to accommodate various use cases. However, using a variety of tools also has disadvantages, such as the complexity of the connections between them and the increase in the number of security management targets. Some experts have harshly criticized, saying, "Even if you use 20 types of generative AI tools, it cannot be called innovative. Rather, it will only lead to a lack of control."
Additionally, there are also cases of generative AI projects at financial institutions that are "dormant." This suggests that there are projects that started in the early experimental stages but have not progressed to full-scale development due to the difficulty of complying with regulations or not achieving the expected results. Another issue is what to do with these projects going forward (restructure them or formally terminate them).
Industry initiatives and expert perspectives: Why are approaches different?
The differences in how the retail and financial industries approach generative AI are rooted in the characteristics of each industry, the environment they operate in, and their "engineering cultures."
Jason Andersen, a principal analyst at a research firm, commented on the average age of generative AI repositories in the retail industry, 453 days, saying, "That's longer than I thought. I thought it would be shorter." This is because, in his opinion, he thought the retail industry had only recently started to get serious about it. On the other hand, he said that the average repository age of 688 days (about two years ago) in the financial industry makes sense, as it coincides with when many early generative AI models began to appear. The financial industry has historically been good at using data and tends to respond quickly to new technologies.
According to Andersen, the financial industry has deep financial resources and a culture of "trial and error with beta (prototypes)," so it is easy to take an experimental approach to new technologies such as generative AI. They tend to view generative AI from the perspective of innovation: "How can we create new financial products and services?"
In contrast, Andersen points out that retail IT departments have traditionally approached automation technologies from the perspective of, "How can this improve our profit margins?" Generative AI is an extension of that same approach, with a focus on the pragmatic aspects of how it can improve customer satisfaction, increase sales and reduce costs.
A report from AI security vendor Apiiro also explains the difference in terms of the nature of the "tasks." Retail teams use generative AI for real-time customer-facing features like recommendation engines and automated support, which have short feedback loops and a direct impact on revenue, so there is always an incentive to "get to market quickly." Meanwhile, financial institutions are subject to heavier regulatory scrutiny, and generative AI is often limited to internal systems or training scenarios that are isolated from production data.
Maman Ibrahim, principal partner at consulting firm Eugene Zonda, made a harsh but apt comment about the diversity of tools in the financial industry: "Introducing 20 different generative AI tools does not mean innovation. It just creates an uncontrollable situation." He points out that the pursuit of flexibility can lead to complex management and increased risks.
The opinions of these experts also highlight the differences in the generative AI strategies of the retail industry, which focuses on speed and practicality, and the financial industry, which focuses on caution and innovation.
Specific use cases and future prospects: How will our lives change?
So how exactly will generative AI be used in the retail and financial industries, and how will it change our future?
Retail Use Cases and the Future
- Personalized Shopping Experience:
- Current: When you visit the online store, AI analyzes your browsing history and preferences to suggest the best products for you in the "Recommended for You" section. Chatbots answer questions about sizes and stock availability in real time.
- future: In virtual fitting, an avatar that looks just like you, generated by AI, will wear the clothes. In stores, smart mirrors equipped with AI will suggest outfits that suit you. We may see the appearance of "AI shopping concierges" that can instantly create the best product list and purchase plan just by telling AI what you want.
- Efficient store operations and supply chain:
- Current: AI analyzes past sales data, weather, event information, etc. to forecast demand several weeks in advance, maintaining optimal inventory levels and preventing stockouts and surpluses.
- future: AI will capture real-time consumer trends and automatically optimize the entire supply chain from manufacturing to delivery. Unmanned delivery by drones and self-driving cars will become commonplace, and products will arrive within a few hours of ordering.
- Engaging Marketing and Advertising:
- Current: AI generates multiple ad copy and designs that resonate with the target customer base, maximizing effectiveness through AB testing.
- future: AI generates completely custom ad content (including videos and music) in real time based on individual tastes and preferences, making it feel like the ads you receive are made just for you.
Use cases and future of the financial industry
- Advanced financial advice and investment management:
- Current: The robo-advisor simply asks a few questions, and AI proposes a portfolio. Within financial institutions, AI analyzes huge amounts of market news and reports, supporting the work of analysts.
- future: AI will design a detailed, lifelong financial plan tailored to your life events (marriage, childbirth, retirement, etc.) and values. AI will also explain the advantages and disadvantages of complex financial products in an easy-to-understand way. However, the final decision will be made by humans, and AI will mainly play the role of a powerful supporter.
- Enhanced fraud detection and security:
- Current: AI monitors 24 hours a day for suspicious patterns of credit card fraud and money laundering.
- future: Generative AI can predict new fraud methods and propose preventative measures. Combined with biometric authentication, more robust and easy-to-use identity verification systems may become widespread.
- Developing innovative financial products and services:
- Current: New indices that utilize AI and investment trusts specializing in specific themes have emerged.
- future: It is possible that services will emerge in which AI can "generate" financial products on demand, based on an individual's risk tolerance and target return. For example, a product that meets specific needs, such as an investment portfolio for a person in their 30s who is interested in environmental issues and seeks stable returns, may be instantly created.
It is expected that generative AI will expand its scope of application in the retail industry, where it will be closer to customers, and in the financial industry, where it will be used in more fundamental and specialized areas.
Retail vs. Finance: A Deep Dive into Generative AI Strategies
As we have seen, there are significant differences in the approaches to generative AI in the retail and financial industries. Let's take a look at the strategies of the two industries and explain what is behind them.
| Comparison points | Retail | Financial industry |
|---|---|---|
| Development speed and production launch | fast.Actively promoting production adoption (e.g., 61% of repositories are in active development) | Slow and careful.Long experimental phase with gradual rollout (e.g. 22% of repositories in active development) |
| Main purpose/focus | Improved customer experience, increased sales, reduced costs, etc.Direct Revenue ImpactReal-time customer response capabilities. | Internal business efficiency, risk management,Innovation and New Product Development(Long-term perspective). Use in internal systems and data-abstraction scenarios will take precedence. |
| data access | Direct access to real-time customer dataThere are many systems that do this (e.g. recommendation engines). | Access to customer data is moreSiloizationVery cautious, with few direct connections to live user data in production. |
| Regulations and risk tolerance | Relatively low (compared to the financial industry). However, personal information protection is important. | Very high.Strict regulations and compliance are top priority. |
| Tool stack (technology used) | A select few.OpenAI Python SDK, LiteLLM, etc. are mainstream. Emphasis on rapid start of operation. | Diverse.We are experimentally using a wide range of tools, including OpenAI Client, LangChain, and LiteLLM. We value flexibility, but there is also a risk of complexity. |
| Development time (average repository age) | Approximately 453 days. Aim for results in a relatively short period of time. | Approximately 688 days. This includes research and development from a long-term perspective. |
| Key points for risk management | Data mapping, access control auditing, early static analysis (find code issues early). | Find sensitive information (such as passwords), organize dependencies, and review dormant projects. |
As can be seen from this comparison, the retail industry is trying to respond quickly to changes in the market by "offensively using AI," while the financial industry is taking a strategy of "strengthening defenses while looking for opportunities to innovate using AI." It's not a question of which is better or worse, but rather the result of each industry choosing the optimal approach to suit its characteristics and objectives.
Risks and Cautions: The Positives and Negatives of Introducing Generative AI
Generative AI is an incredibly powerful technology, but its deployment also comes with some risks and caveats, and understanding these is essential to maximizing the benefits of the technology and avoiding problems before they arise.
General Generative AI Risks
- Hallucination (plausible lie): AI can confidently generate factually unfounded or even incorrect information, and if we take this information at face value, we could make a big mistake.
- Bias: AI can reflect the biases contained in the training data, which can lead to the risk of making unfair judgments against people with certain attributes.
- Information leakage and security: If confidential or personal information is handled improperly during the training or operation of AI models, it could lead to information leaks. In addition, the AI system itself could be subject to malicious attacks.
- Copyright and Intellectual Property Rights: Legal issues are also being debated, such as who owns the copyright to content generated by AI and whether AI infringes copyright on content used as training data.
- Potential Exploits: There is also concern that it could be used for malicious purposes, such as generating fake news, creating fraudulent emails, and generating malware (malicious program) code.
Risks and precautions specific to the retail industry
- Customer Data Handling: Because we handle large amounts of customer data for personalization, we need to be extremely careful about privacy protection and data security. If a data breach occurs, it will seriously damage customer trust and your brand image.
- Customer churn due to incorrect recommendations: If AI-generated product recommendations are off-topic or off-putting, customers will abandon them.
- Incorrect chatbot responses: If an AI chatbot provides inaccurate information or responds rudely to customer inquiries, it could damage a company's reputation.
Apiiro advises the retail industry that "data mapping (understanding what data is where), access control audits, and static code analysis (a method to detect program problems early) are important in the early stages of development to detect problems before they are deployed to production environments."
Risks and points of caution specific to the financial industry
- The Complexity of Regulatory Compliance: The financial industry is particularly regulated, and when introducing new technology, careful checks must be made to ensure that all relevant laws and regulations are met. Non-compliance can lead to large fines and business suspension orders.
- The serious impact of system errors: Misjudgments by AI in financial trading or risk management systems could not only cause direct losses to customer assets, but could even cause chaos across the entire market.
- Ensuring accountability: When an AI makes a decision (such as whether or not to grant a loan), humans need to be able to understand and explain why it has reached that conclusion. This is also linked to research in the field of "Explainable AI (XAI)."
- Declining governance due to proliferation of tools: As mentioned above, introducing many tools on an experimental basis runs the risk of complicating the security management and integration of each tool, making it difficult to maintain control across the organization.
- Managing dormant projects: AI projects that are left abandoned for long periods of time may develop security holes or remain outdated technology, so they require regular review and appropriate disposal (restructuring or retirement).
Apiiro recommends that the financial services industry prioritize "detecting sensitive information (such as API keys and passwords), maintaining the health of software dependencies, and reevaluating dormant generative AI projects to see whether they should be rebuilt or retired."
Properly managing these risks and paying careful attention to ethical considerations will be key to popularizing generative AI in a way that benefits society.
Current Trends and Future Directions
Technological innovations surrounding generative AI are evolving day by day, and both the retail and financial industries are taking various steps to keep up with this new wave. Let's take a look at current trends and what the future holds for the next few years.
Retail trends and the future
In retail, generative AI"Faster and deeper"The trend of incorporating it into customer experiences is accelerating. In particular,"Omnichannel experience"Generative AI will play a central role in making this a reality. For example, a seamless purchasing experience is possible, where a customer chooses a product online while consulting with AI, and then makes a final decision after watching an AI avatar try it on in a physical store. In addition, back-end operations will become more efficient and sophisticated, including supply chain optimization and hyper-personalized dynamic pricing (which changes prices according to demand).
Trends and the future of the financial industry
The financial industry remains cautious, but"From Experiment to Implementation"We are gradually beginning to see the emergence of AI. In particular, we expect to see the practical application of AI in areas such as business support tools for bank employees (such as information search and report creation support by AI assistants), risk management, and compliance checks. As for customer services, more advanced personal finance management (PFM) tools and fraud prevention systems will become more widespread. In the long term, we also expect to see the development of new financial derivatives using generative AI and the construction of more precise market prediction models, but this will require close cooperation with regulators and the formation of a social consensus.
Future directions common to both industries
- Establishing AI ethics and governance: It will become increasingly important to ensure the transparency, fairness, and accountability of AI decisions. Industry groups and regulatory authorities will likely begin to formulate guidelines.
- Talent development and reskilling: To master generative AI and maximize its benefits, it is imperative to develop human resources with knowledge and skills related to AI. Reskilling of existing employees will also be important.
- Start small and continue to improve: Rather than aiming for a perfect system from the start, the mainstream approach will be to start with a small-scale introduction and then gradually expand it while verifying its effectiveness.
- Human and AI collaboration: Rather than having AI replace all work, emphasis will be placed on a "collaborative model" that combines areas in which humans excel (empathy, creativity, complex decision-making, etc.) with areas in which AI excels (mass data processing, pattern recognition, etc.) to utilize the strengths of each.
Generative AI is becoming an essential technology in shaping the future of retail and finance. The speed of its evolution is remarkable, and in a few years we may see the emergence of services that we can't even imagine now.
Frequently Asked Questions (FAQ)
- Q1: What exactly can generative AI do?
- A1: Generative AI is an AI that can learn from large amounts of data and "create" new text, images, sounds, program codes, etc. For example, it can answer questions in natural language, write stories, come up with design ideas, summarize complex data, and do a very wide range of things. It is beginning to be used in the retail industry to suggest products that suit customers, and in the financial industry for market analysis.
- Q2: What are the benefits of using generative AI in the retail industry?
- A2: The biggest advantage is that we can provide personalized services for each customer."Personalized experiences"For example, you can see special recommended products just for you, and an AI chatbot will respond to you kindly 24 hours a day. This will improve customer satisfaction and lead to increased sales. In addition, business efficiency can be greatly improved by automatically creating product descriptions and forecasting demand.
- Q3: Is generative AI safe in the financial industry?
- A3: The financial industry handles valuable assets of customers, so the introduction of generative AIVery carefulSafety, reliability, and compliance with laws and rules are given top priority. Therefore, thorough testing and verification are carried out to ensure that AI does not make incorrect decisions, and in many cases, a system is in place for humans to check the AI's decisions. Technology is advancing every day, and the development of safer and more reliable systems is progressing.
- Q4: Why is the use of generative AI different in retail and finance?
- A4: That's because each industry has different issues, goals, and rules to follow. The retail industry is all about how to satisfy customers and get them to buy our products.Speed and customer experienceOn the other hand, the financial industry is focused on protecting customer assets and not making mistakes."Safety, reliability and regulatory compliance"That's why, while retailers are quick to try out new customer-facing features, finance companies tend to start with more cautious implementations of less visible features, such as improving the efficiency of internal operations and risk management.
Summary and Related Information
In this article, we have explained how generative AI is being used in the retail and financial industries and how the approaches differ in each industry. The retail industry is quickly adopting AI to improve customer experience and increase revenue, while the financial industry is pursuing innovation with a careful and long-term perspective while taking into account regulations and risks. There is no doubt that both industries are entering an era of great change due to generative AI.
This technology is still in its infancy and many challenges remain, but its possibilities are endless. Let's continue to keep an eye on the evolution of generative AI and see how it will change our lives and society!
Related links (tips for gathering information)
- Ministry of Economy, Trade and Industry "Information on AI": For those who want to know about Japan's AI policies and related information.
- Japan Deep Learning Association (JDLA): If you want to know the latest trends in AI technology and qualification information.
- Industry news sites (e.g. Nikkei Crosstech, ZDNet Japan, etc.): When searching for specific company case studies or technical commentary articles.
- Corporate blogs from OpenAI, Google AI, and others: If you want to know about the latest AI models and research results.
This article does not recommend investing in specific financial products or services, or in the stocks of individual companies. Information on AI technology changes daily, so please check the latest information yourself and make careful decisions.