The Gen AI Productivity Paradox: A Warning for Financial Services

Dear Visionary Leader📣 - Thank you for connecting. We share cutting-edge, practical insights on how to win in the age of AI. If you're new to our network and would like to enhance your AI strategy, click the "subscribe" button above...

This article is 2100 words long and takes about 5 minutes to read (but longer to digest…)

Executive Summary

Charlie Munger, the Vice Chairman of Berkshire Hathaway, famously observed that productivity gains from new technologies in the textiles industry or from robotic automotive assembly lines have primarily benefited customers and suppliers rather than the companies deploying the innovations.

The machinery improved, productivity soared, but the competitive playing field soon levelled off, and profit margins for producers remained thin, with little benefit to shareholders.

This insight offers an important warning for financial institutions – banks, insurers and others - rushing to adopt Generative AI (Gen AI). While financial services are significantly less commoditized than textiles or automobiles - with complex regulatory requirements, high trust barriers, and sophisticated risk management needs - the risk of value leakage remains real.

This edition of AI Risk & Strategy examines how this "productivity paradox" is playing out in financial services and provides a strategic framework for executives to navigate the Gen AI revolution.

In it we cover:

• The Economics of Productivity Gains
• The Looming Threat to incumbents
• Understanding the Value at Risk
• The New Market Reality
• Building Sustainable Advantage
• Taking Action: A CEO's Roadmap

The Economics of Productivity Gains

Gen AI's impact on financial services productivity is substantial but complex. In operations, it can reduce processing times by up to 80% for routine tasks. In risk assessment, it can analyse vast amounts of unstructured data to improve prediction accuracy. For fraud detection, it can identify patterns that human analysts might miss, potentially saving billions in fraudulent claims.

However, capturing these gains as sustainable profit proves challenging for several reasons. First, the implementation costs are substantial - beyond the direct technology costs, institutions must invest in infrastructure upgrades, talent acquisition, and staff retraining. Second, regulatory requirements around AI use in financial services often necessitate additional compliance investments. Third, as these capabilities become standardized, competitive pressure often forces institutions to pass savings on to customers through lower fees or improved service levels.

Consider the case of a major European bank that invested €50 million in Gen AI implementation. While they achieved a 70% reduction in loan processing time, their profit margins on standard lending products actually decreased as competitors matched their capabilities and price points.

The Looming Threat

Consider JPMorgan Chase's recent implementation of Gen AI for commercial loan documentation. The system can now generate complex loan agreements in minutes rather than days, analyse unstructured customer data for credit decisions, and create personalized term sheets.

While the productivity gains are impressive - 40% reduction in processing time, 50% decrease in documentation errors, and significantly improved customer satisfaction - they've come at significant cost. Beyond the direct technology investment, JPMorgan has had to invest heavily in AI talent acquisition, staff retraining, and infrastructure upgrades. Moreover, regulatory scrutiny of their AI systems has required substantial investment in compliance and risk management.

Yet within months, smaller regional banks began offering similar capabilities through off-the-shelf Gen AI platforms.

The competitive advantage JPMorgan hoped to gain quickly eroded, while the costs of AI licensing continued to rise. This scenario is playing out across the financial services industry, from wealth management to insurance.

Understanding the Value at Risk

The impact of Gen AI varies significantly across different aspects of financial services. To understand where the risks and opportunities lie, we need to examine how Gen AI affects the core value propositions of different financial institutions.

For traditional banks, the foundation of their business – trust, security, and complex financial intermediation – remains relatively protected from Gen AI disruption. However, their standard credit assessment and basic advisory services are highly vulnerable to commoditization. When Anthropic's Claude can analyse a company's financials and generate a detailed credit assessment report in seconds, the traditional bank's advantage in standard lending decisions diminishes significantly.

Insurance companies face a similar challenge. While their fundamental role in risk pooling and capital provision remains secure, their traditional advantage in risk assessment for standard policies is rapidly eroding. When Gen AI can analyse policy documents, assess risk factors, and generate coverage recommendations instantly, the traditional underwriting advantage disappears for all but the most complex risks.

Insurance brokers perhaps face the greatest threat. Their traditional role in price discovery and standard product advisory can be largely automated by Gen AI platforms. However, their value in complex claims advocacy and sophisticated risk advisory remains difficult to replicate with technology alone.

The New Market Reality

The threat comes not just from traditional competitors adopting Gen AI, but from new entrants who can now replicate core financial services capabilities without the burden of legacy systems.

Stripe's new Gen AI-powered lending service, for instance, can now automatically assess creditworthiness and generate loan offers for millions of small businesses, capabilities that were once the exclusive domain of banks. Meanwhile, Lemonade has fundamentally reimagined insurance operations by building their entire technology stack around AI from the ground up.

Their Gen AI platform "Jim" can process claims instantly, customize policies in real-time, and handle customer interactions seamlessly - demonstrating how core insurance processes can be completely reimagined when unburdened by legacy systems and traditional operating models. While Lemonade's profitability remains unproven, their technological approach has forced traditional insurers to rethink what's possible in insurance operations.

The economics of Gen AI implementation increasingly favour three groups:

1. Customers who benefit from improved service and lower costs
2. Technology providers who can leverage their solutions across multiple institutions
3. Skilled employees with AI expertise who command premium compensation.

Financial institutions, meanwhile, often find themselves caught in an arms race of continuous investment just to maintain competitive parity.

Building Sustainable Advantage

Forward-thinking financial institutions are finding ways to turn the Gen AI paradox to their advantage. They focus on four key strategies:

• First, they carefully distinguish between core and complementary applications of Gen AI. Morgan Stanley, for example, uses Gen AI to enhance rather than replace their research analysts, combining the technology's data processing capabilities with their experts' judgment to create unique insights.
• Second, they build proprietary AI advantages by training models on their unique data assets. UBS's wealth management division has developed Gen AI models trained on decades of high-net-worth client interactions, creating capabilities that generic AI models cannot match.
• Third, they shift value creation to protected areas where proprietary AI capabilities create sustainable competitive barriers. Ping An Insurance Group exemplifies this approach. The company has built one of the world's largest AI patent portfolios outside of pure technology companies, with over 33,000 patent applications and 11 consecutive years of leading global financial institutions in fintech patent applications. Their investment goes far beyond simply implementing AI - they've created proprietary AI capabilities in areas ranging from intelligent underwriting to medical imaging diagnosis.

For instance, their Gen AI platform not only processes insurance claims but integrates with their own adjacent healthcare ecosystem to provide real-time risk assessment and personalized health management.

This deep integration of AI into their core business model, backed by extensive proprietary technology, creates competitive advantages that can't be easily replicated by simply licensing third-party AI tools.

• Fourth, they grasp the potential of ‘Agentic AI’, the ability to enrol a new near infinite number of non-human, ‘synthetic’ workers to supplement and complement their human workforce. Financial institutions are just waking up to this opportunity – those who make the boldest move will win big.

Taking Action: A CEO's Roadmap

Start with Strategic Triage

Begin by categorizing your organization's activities into three buckets:

1. Those that Gen AI will likely commoditize
2. Those it will enhance
3. Those it may transform

This isn't a one-time exercise. AXA, for example, maintains a quarterly review process where senior executives reassess each major business function against emerging Gen AI capabilities.

For activities facing commoditization – such as standard loan documentation or basic insurance policy issuance – the goal should be cost reduction while transitioning to higher-value services. Activities that Gen AI can enhance – like complex risk assessment or wealth management advisory – should receive focused investment to build proprietary capabilities. For potentially transformative areas – such as personalized financial planning or real-time risk pricing – the priority should be experimentation and learning.

Build Your Moat

Traditional competitive moats in financial services – scale, brand, distribution networks, consumer apathy – while still important, may not provide sufficient protection in the Gen AI era. New moats need to be built around three key elements:

1. First, proprietary data assets must be organized and enriched. Standard Chartered Bank, for example, is creating unique value by combining transaction data with Gen AI to provide market-specific insights that generic models cannot match.
2. Second, develop unique process knowledge that Gen AI can enhance but not replace. Goldman Sachs is building proprietary Gen AI models that incorporate decades of deal-making expertise, creating capabilities that go beyond what off-the-shelf AI systems can provide.
3. Third, focus on complex, judgment-intensive activities where human expertise and AI can create synergistic value. These are areas where the combination of institutional knowledge and AI capabilities creates outcomes that neither humans nor AI could achieve alone.

Transform the Operating Model

Success with Gen AI requires more than just technological investment – it demands a transformation of how financial institutions operate. A clear strategy and new operating model is required to deliver sustainable value creation. The operating model needs to take account of business purpose, risk profile and talent.

The talent challenge is particularly crucial. Rather than simply competing for scarce AI specialists at premium salaries, successful institutions are focusing on building hybrid teams that combine domain expertise with AI literacy and - expect to see this discussed a lot in 2025 – hybrid teams of humans and AI agents. These approaches not only reduce dependency on expensive AI talent but create more sustainable capabilities that leverage institutional knowledge.

This transformation requires:

• First, rethinking talent strategy: Build teams that combine financial expertise with AI literacy. This doesn't mean every banker needs to become a data scientist, but they need to understand how to work effectively with AI systems.
• Second, redesigning processes: Create workflows that optimize the human-AI partnership rather than simply automating existing processes. For example, Allianz is redesigning their commercial insurance underwriting process to leverage Gen AI for data analysis while focusing human underwriters on complex decision-making and client relationships.
• Third, reimagining customer relationships: Use Gen AI to enhance rather than replace human interactions. UBS's wealth management division, for instance, uses Gen AI to provide their advisors with deeper insights while maintaining the crucial human element in client relationships.
• Fourth, appreciate the imminence of Agentic AI: a new ‘synthetic’ workforce of AI agents will soon be able to collaborate in teams with human workers. This dramatically changes the economics of firms. For example, one small insurance company is running an entire operations department for a new product line with AI agents.

Manage the Transition

The shift to Gen AI-enabled financial services must be managed carefully to maintain stability and trust. We recommend a three-horizon approach, one updated for the AI Age:

• Horizon 1: Focus on protecting core value propositions while building foundational AI capabilities. Prioritize quick wins in areas like document processing and customer service, but avoid over-investing in easily commoditized capabilities.
• Horizon 2: Develop proprietary AI capabilities in protected areas. This includes building unique data assets, training specialized AI models, and creating integrated human-AI workflows.
• Horizon 3: Transform the business model to capitalize on new opportunities created by Gen AI while maintaining competitive advantages in core areas.

Whereas in the past you might address these horizons sequentially over two or three years, in the Age of AI, given how fast things are moving, we recommend addressing them in parallel, as part of an approach we call ‘Triple Transformation’.

Conclusion

The Gen AI productivity paradox in financial services isn't inevitable, but avoiding it requires a sophisticated understanding of where and how value is created and captured.

Success demands more than just implementing Gen AI - it requires strategic choices about where to invest, how to build sustainable advantages, and how to retain the value created rather than seeing it competed away or captured by others.

The winners in this transformation will not be those who adopt Gen AI most aggressively, but those who deploy it most strategically, especially Agentic AI. They will focus on protected spaces where human judgment remains crucial, build new moats around proprietary data and expertise, and transform their operating models to optimize human-AI collaboration.

The time to act is now, but the action must be strategic rather than reactive. Financial institutions that follow this approach will be well-positioned to thrive in the age of Gen AI, turning the productivity paradox from a threat into an opportunity for sustainable competitive advantage.

The lesson from Charlie Munger's textile mill story, while not perfectly analogous to financial services, remains relevant: in the face of technological change, strategic wisdom often means knowing where not to compete as much as where to invest. In the age of Gen AI, this wisdom will separate the winners from those who chase productivity gains at the expense of sustainable advantage.

--------------------------------------------------------------------------

Please share this edition of the AI Risk & Strategy newsletter with others and encourage them to subscribe, for free, at this link. You can also follow us on LinkedIn at AI Risk