<div id="readability-page-1">Banks will leverage Explainable AI (XAI) tools like SHAP and LIME to demystify complex models, making AI-driven decisions and recommendations more transparent for users. getty A report by the Chartered Financial Analyst Institute released this fall underscores a growing tension inside financial services. While firms are rushing to deploy AI, few are doing the work needed to make those systems transparent and trustworthy. In Explainable AI in Finance: Addressing the Needs of Diverse Stakeholders, the institute warns that opaque decision making in credit scoring, underwriting, fraud detection, insurance, and investing threatens consumer trust, regulatory compliance and institutional risk.  To understand how financial institutions are navigating this tension on the ground, I spoke to Gill Haus, Chief Information Officer of the Consumer Community Bank at JPMorgan Chase &amp; Co., and Marianna Lopert-Schaye, Global Head of Strategic Partnerships, Investments, and Innovation Centres at BNY.  For more like this on Forbes: Precision, Not Hype, Will Shape Banks’ Use Of AI In 2026 .  Video: Gill Haus, Chief Information Officer of the Consumer Community Bank at JPMorgan Chase &amp; Co  MORE FOR YOU From Experimentation to “AI for Everyone” While JPMorgan Chase focuses on massive scale and efficiency, BNY is aggressively moving to democratize access to these tools across its organization.  &#34;The goal is really moving beyond experimentation,&#34; says Lopert-Schaye. BNY’s mandate is to deploy “AI for everyone, everywhere, for everything.”  BNY now has over 120 AI solutions in production, driving everything from predictive analytics and anomaly detection to operational efficiencies. The bank uses a platform called Eliza, a model-agnostic infrastructure that allows the firm to interact with various AI models while maintaining strict compliance and security guardrails.  For more like this on Forbes: What Is Agentic AI And What Will It Mean For Financial Services?  Video: Marianna Lopert-Schaye, Global Head of Strategic Partnerships, Investments, and Innovation Centres at BNY.  The Trust Gap: Why &#34;Because the AI Said So&#34; Isn’t Enough However, scaling AI in a regulated industry like finance faces the unique hurdle defined in the CFA Institute report: opacity. The &#34;black box problem&#34; makes it difficult to trust an AI system’s decisions, assess model fairness, and meet regulatory demands.  You cannot deny a loan, flag a transaction for fraud, or rebalance a pension fund without being able to explain why. The CFA report highlights that deep learning models require explainability to ensure fair lending decisions, as they can incorporate alternative data sources that might inadvertently proxy for protected attributes like ethnicity or gender.  &#34;If X, then you do Y... technologies in the past were a bit more deterministic,&#34; Haus observes. But with Generative AI, &#34;it might give you a slightly different answer.&#34; To bridge this gap, banks are turning to XAI (Explainable AI), capabilities seeking to provide human-understandable justifications for AI output.  According to the CFA Institute, XAI is becoming essential for regulatory compliance and risk management. Techniques like SHAP (Shapley Additive Explanations) allow analysts to see exactly which features, such as a debt-to-income ratio or a specific transaction history, influenced a model’s decision.  For instance, if a credit model denies a loan, an &#34;explanation by simplification&#34; method like LIME can approximate the complex model's behavior to tell the customer: &#34;If your income were $5,000 higher, the loan would be approved.&#34; This moves AI from a mysterious oracle to a transparent tool.  The Human-in-the-Loop: A Feature, Not a Bug Despite the advanced automation, both executives emphasize that humans remain central to the process. The fear of AI replacing bankers is being replaced by the reality of AI augmenting them.  &#34;We don’t let it loose in that sense,&#34; Haus says of Generative AI. &#34;We have a human in the middle.&#34; At JPMorgan Chase, AI might transcribe a customer call and recommend an answer, but a human agent decides whether to use it. &#34;The human being that's sitting there talking on the phone is able to intermediate,&#34; Haus adds.  At BNY, the focus is on &#34;freeing up lots of people's time for higher value work.&#34; Lopert-Schaye notes that nearly 98% of the firm's 55,000 employees have completed responsible AI training. The cultural change, getting people comfortable with a new way of working, is often &#34;as important, if not more important, than some of the technical... change.&#34;  Strategic Patience: The Long Game For fintech startups looking to work with these giants, or executives looking to implement similar strategies, the lesson is patience and alignment. Innovation in a global bank doesn’t happen overnight.  Lopert-Schaye reveals that the cycle from testing a new technology to full deployment can take 12 to 18 months. Success requires identifying a &#34;champion&#34; within the organization, a respected subject matter expert who can advocate for the technology. &#34;Don't think about non-technical colleagues as being a challenge,&#34; she advises. “Think of them as being an opportunity to bring someone on the journey.”  Haus agrees, noting that successful executives in this space are those who obsess over the customer problem, not the tech stack. &#34;It is technology for solving an actual business problem,&#34; he says. “If you understand that, then you're better able to apply technology.”  The Road Ahead As financial institutions look to the next 12 months, the focus remains on &#34;measurable business outcomes.&#34; Whether it is acquiring new capabilities to drive growth or enabling operating efficiencies, the era of AI tourism is over.  In this high-stakes environment, AI will continue to grow in influence, but only as fast as it can be explained. As the industry evolves, the future belongs to &#34;neurosymbolic&#34; and hybrid models that balance the raw power of deep learning with the logical transparency required by law and ethics.  Follow Holloman to learn more about the future finance and technology.  Editorial Standards Reprints &amp; Permissions</div>

摩根大通

<p>大型銀行計劃到 2026 年通過使用可解釋人工智能（XAI）工具，如 SHAP 和 LIME，使人工智能變得更加可訪問和可解釋。這旨在增強人工智能驅動決策的透明度，以解決消費者信任和合規性問題。摩根大通和 BNY 正在領導使人工智能訪問民主化的努力，強調人類在人工智能過程中的參與。CFA 協會強調了人工智能可解釋性的重要性，以確保公平貸款和遵守監管規定。對於與銀行合作的金融科技初創公司來説，戰略耐心至關重要</p>

大型銀行計劃在 2026 年如何讓人工智能變得更易於訪問和解釋