The entry of ChatGPT into the technosphere in late 2022 sparked a global frenzy over the possibilities embedded in generative AI (GenAI). For the first time, non-technologists could feel the pulse of artificial intelligence at their fingertips. From answering questions and generating images to writing code, GenAI captivated the public imagination — while C-suites scrambled to translate that buzz into bottom-line value.​

According to McKinsey, generative AI could contribute between $2.6 and $4.4 trillion annually to the global economy, with $200 to $340 billion accruing to financial services alone. Governments and industries around the world have since launched into a race to capture this potential windfall. But behind the compelling figures and projections of economic growth lies a critical question: Who will benefit?

Growth without purpose

We find ourselves at a stage in global development where the goal and end product of economic activity is compounding growth on top of preexisting growth. That is to say, the pursuit of growth has become the end in itself without consideration of the purpose beyond that end. That single-minded pursuit has obscured the pressing question of which parts of society remain devoid of basic financial security. If the financial sector were to take that question as its starting point, generative AI could serve as a powerful lever for closing financial gaps — rather than merely reinforcing the advantages of existing winners.

However, this deeper sense of why has largely been absent from the financial sector’s approach to innovation and growth. Like many industries, it has primarily viewed generative AI through the narrow lens of operational efficiency and productivity gains. Most incumbents have positioned generative AI as a cost-saving tool — auto-drafting research notes, expediting KYC checks, and triaging customer service interactions.

Prevalent use cases of generative AI in financial services; Graphic by Clarisse Awamengwi

The use cases focused on productivity and efficiency are not without merit. In fact, these gains can contribute to financial inclusion—albeit indirectly. By lowering the operational costs of banks and financial institutions, they create opportunities to serve lower-income segments that were previously deemed unprofitable. However, this alone is not sufficient.

The real opportunity lies in shifting from productivity to purpose. Generative AI has the potential to dismantle knowledge barriers, customize financial products for underserved microsegments, and direct capital toward historically excluded communities. But realizing that potential requires intentionality. Banks and financial institutions must actively leverage generative AI — and adjacent technologies — to create societal value and expand meaningful access to financial services. Startups like Cleo AI offer a glimpse of what purposeful design can achieve. Incumbent banks, with their balance sheet strength and market reach, have the capacity to scale that impact — if they choose to innovate with intention.

Innovation must always answer two essential questions: For whom? And for what purpose?

Two recent pilots illustrate this untapped potential. Morgan Stanley’s “Debrief” generates and summarizes client notes for financial advisors. Royal Bank of Canada’s “North for Banking” supports its capital markets advisors with insights and analyses. Useful applications, yes, but inwardfacing and non-expansive in scope and reach. Imagine if the same models were pointed outward — answering the questions firsttime investors have in plain language, in Hindi and Swahili as readily as in English.

According to the 2021 World Bank Findex Report, 1.4 billion adults around the world remain unbanked. Moreover, of the 1.1 billion adults who have entered the formal financial system — through banks or mobile money providers — since 2011, the majority remain underbanked. The services they receive often do little to improve their financial well-being. A basic account that fails to enhance day-to-day security or long-term prospects does not constitute meaningful financial inclusion. Closing the gap from unbanked to well-banked requires financial institutions to harness generative AI as a catalyst for access, education, wealth-building, and everyday financial management — not merely as a cost-cutting tool. This shift demands a mindset change: from viewing generative AI as a means of boosting productivity to recognizing it as a gateway to genuine financial inclusion.

Proposed financial inclusion use cases of generative AI; Graphic by Clarisse Awamengwi

Levers to Bank the Un(der)banked

1. Financial Education & Literacy

Generative AI’s chat-first interface can serve as a low-cost, personalized tutor for individuals who have never received formal financial education. By responding to questions in accessible language, offering context-aware tips, and translating community-based savings practices—such as njangi — into bankable product concepts, it enables a two-way learning loop: consumers gain confidence, while institutions uncover unmet needs they can design for.

Proof of concept: Ally Financial’s “Bank On Me” pilot combines a GenAI-powered bot with gamified lessons that nudge users to build emergency savings or pay down high-interest debt. Early testers reported a 28% increase in weekly engagement compared to Ally’s static FAQ pages.

2. WealthBuilding Pathways

Knowing about money is not the same as growing it. Racial, gender, and class barriers continue to shape who invests, who benefits from compounding returns, and who is left holding cash. Even the so-called “democratization” of markets through zero-commission apps has its limits — few first-time investors are likely to swipe to buy if no one explains the risks, the jargon, or the basics of portfolio construction.

Here, generative AI can act as a virtual wealth coach: breaking down market news into plain language, modeling long-term financial scenarios, and nudging users toward diversified, goal-aligned portfolios. By combining real-time insights with bite-sized learning, it transforms abstract exposure into actionable participation — helping underserved communities shift from underbanked to asset-building.

Proof of concept: JPMorgan’s headline-grabbing “IndexGPT” prototype demonstrates the technical feasibility of GenAI-generated portfolios. A public-benefit twist would be to train that same engine to accommodate micro-investment thresholds and ESG or Sharia-compliant screens tailored to emerging market savers.

3. Technological Accessibility

Technology that excludes is no bridge at all. The mobile money revolution — sparked by Kenya’s M-Pesa — proved that when tools meet people where they are, adoption can leapfrog traditional infrastructure. Today, generative AI offers the potential to dismantle two long-standing barriers.

Language: Most banking apps still default to colonial or elite languages. Training models in Swahili, Hausa, or Pidgin — an effort already underway by the Masakhane NLP community — allows users to engage with chatbots in the languages they actually speak.

Disability: Multimodal generative AI (text, speech, image) can read account statements aloud, convert voice commands into secure transactions, and guide visually impaired users through unfamiliar fintech interfaces. As an in-app co-pilot explaining each step, GenAI can link knowledge to action and transform access into enduring inclusion.

Generative AI has opened a vast new design space; whether it becomes a force for empowerment or for entrenchment depends on the choices we make now.

Proof of concept: India’s government-backed Bhashini initiative offers a glimpse of this at scale — open-source language models supporting 22 official languages and 121 dialects. Financial institutions that integrate with such platforms could reach hundreds of millions who have never encountered a disclosure form in their mother tongue.

4. Innovating with Intention

Generative AI has opened a vast new design space; whether it becomes a force for empowerment or for entrenchment depends on the choices we make now. To ensure the technology bends toward inclusion, financial institutions, startups, and regulators can take five concrete steps:

1. Set inclusion KPIs alongside efficiency targets when funding GenAI pilots and evaluating return on investment.
2. Co-create with underserved communities — from languagemodel finetuning to userexperience testing — so products reflect lived realities, not boardroom assumptions.
3. Invest in language and accessibility research, expanding model coverage of non-European languages and multimodal interfaces that serve users with disabilities.
4. Scale proven models through partnership: pair impactdriven fintech innovators with incumbent banks that command distribution and balancesheet muscle.
5. Embed safeguards early — robust data governance, bias audits, and transparent model cards — to keep trust from becoming an afterthought.

From outputs to outcomes

Outcomes — not outputs — must anchor this agenda. It’s easy to celebrate the launch of a flashy chatbot; far more meaningful is measuring how many users build three months of emergency savings, or graduate from having no account to owning a wealth-building investment portfolio.

Innovation must always answer two essential questions: For whom? And for what purpose?

If generative AI can expand access to knowledge, capital, and opportunity, it will fulfill its highest potential — propelling the journey from unbanked to well-banked, and ultimately, to financially empowered citizenship.