Skip to content

AI and Financial Decision-Making: Who Is Really Making the Decisions?

Artificial intelligence is changing how people access financial information, evaluate opportunities and make decisions. It offers new possibilities for financial education, accessibility and decision support. But as AI becomes a more influential participant in personal finance, an important question emerges: are we becoming better financial decision-makers, or simply becoming more dependent on the systems helping us decide?

1. The new financial intermediary

Imagine wanting to understand whether to pay off a loan early, increase your pension contributions, invest in an index fund or save towards a property deposit.

Until relatively recently, finding answers might have involved speaking to a financial professional, consulting a bank, reading financial publications, searching online or asking someone with experience.

Today, a conversational AI tool can help explain the difference between financial products, summarise investment concepts, compare scenarios and organise questions that might otherwise feel overwhelming.

The change is significant. Financial information is no longer accessed only through traditional institutions, specialist publications or professional intermediaries. It can increasingly be explored through a conversational interface that responds to individual questions.

People can ask questions in their own language, request simpler explanations, revisit unfamiliar concepts and explore financial decisions at their own pace. They can also use AI to organise information before approaching a financial institution or professional.

But this development changes more than the way financial information is accessed. It changes the relationship between information and judgement.

AI can explain, interpret, compare and generate recommendations. Depending on the system and how it is used, it may influence the decisions people ultimately make.

The question is no longer only where we obtain financial information. It is how we evaluate it, how much authority we give it and whether we retain the ability to make informed decisions.

2. From information to decision-making

There is an important distinction between accessing financial information and making a financial decision.

Information helps us understand a subject. Decision-making requires us to consider our circumstances, objectives, constraints, risks and alternatives.

For example, an AI tool can explain the difference between fixed and variable interest rates. It can calculate the impact of different repayment amounts or help compare mortgage scenarios.

But deciding which mortgage is appropriate requires more than understanding interest rates. It involves considering income stability, future mobility, existing obligations, liquidity needs, housing costs and the possibility that personal circumstances may change.

Similarly, an AI system can explain diversification and illustrate how different asset allocations behave under hypothetical market conditions. That does not automatically establish which investment strategy is suitable for a particular person.

It is useful to distinguish four functions:

  • Information: explaining financial concepts, terminology and processes.
  • Education: helping people develop an understanding of financial principles.
  • Decision support: organising information, comparing alternatives and exploring scenarios.
  • Advice: recommending a course of action in light of an individual’s circumstances.

These functions are related, but they are not interchangeable. The fact that a tool can explain a financial product does not mean it has assessed whether that product is suitable. Nor does a personalised-sounding response necessarily constitute regulated financial advice.

Understanding this distinction is becoming an important part of financial literacy.

3. The opportunity: access, affordability and the freedom to ask

One of AI’s most valuable contributions may be its ability to make complex financial information easier to approach.

Financial systems can be difficult to navigate. Investment terminology, pension rules, taxation, credit products, insurance policies and property financing involve technical language and institutional processes.

But there is another barrier to financial understanding that deserves attention: the discomfort of asking questions.

Money is not simply an economic subject. It is also deeply connected to identity, independence, status, family expectations and personal relationships. Financial difficulties can generate shame. Not knowing how a financial product works can create embarrassment. Discussing debt, low income, insufficient savings or investment losses may expose vulnerabilities that people would rather keep private.

Research on consumer debt has demonstrated that anticipated social judgement can lead people to conceal their financial circumstances and avoid seeking help. In a series of studies, Moorhouse and colleagues found that anticipated stigma around debt was associated with concealment behaviours, including avoiding help, with negative consequences for debt reduction and wellbeing.[1]

More recently, StepChange’s October 2026 research among more than 1,600 debt-management-plan clients found that embarrassment was the most commonly reported reason for delaying debt advice. Nearly one in five clients said embarrassment had prevented them from getting help.[2]

These findings concern debt and help-seeking, not the use of AI. Nevertheless, they illustrate why the social environment in which financial questions are asked matters.

Conversational AI introduces a different interaction.

People can ask questions without having to disclose their uncertainty to another person. They can repeat a question, request a simpler explanation, explore an unfamiliar subject or admit that they do not understand a financial concept.

There is no human listener whose reaction they have to manage. The interaction can feel less exposing and less intimidating.

This ability to ask without embarrassment is potentially significant. It may lower some of the social and psychological barriers to financial learning.

There is, however, an important distinction between the privacy of an interaction and data privacy. A conversation may feel private because there is no human listener, but the information entered into an AI service is still subject to the provider’s data practices, settings and privacy policies. Users should be cautious about sharing identifying details, account numbers, passwords or sensitive financial records.

The economics of access

There is also an economic dimension.

Some AI tools offer free access or relatively inexpensive subscriptions. For people who cannot afford regular consultations with financial professionals, they may provide a lower-cost starting point for understanding financial concepts, organising questions and exploring financial scenarios.

The potential affordability of AI-supported financial guidance is increasingly part of the research conversation. A 2026 study by Baeckström and Matkovskyy in the Journal of Corporate Finance compares recommendations from professional human advisers and generative AI systems. The researchers identify the potential for AI-powered co-advisers to deliver more scalable and affordable guidance. However, their experiment also finds meaningful differences in portfolio recommendations and modelled investment outcomes. The cost of human advice matters when comparing the two approaches, but the study does not establish that AI and professional advice are interchangeable.[3]

AI can therefore offer a relatively inexpensive first point of support, but affordability should not be confused with suitability or professional accountability.

Complex decisions involving taxation, legal obligations, regulated investment products, pensions or substantial assets may require qualified expertise and an assessment of individual circumstances.

Nevertheless, the cost of accessing financial knowledge matters. If people can explore basic financial questions at a lower cost, they may be better positioned to understand their options and prepare for more specialised advice.

The opportunity is not simply to make financial information available. It is to make asking, learning and exploring financial decisions more accessible.

But accessibility is only the beginning. The ability to understand and evaluate the information remains essential.

4. The risks behind the answer

AI systems can generate inaccurate information, misunderstand a question, rely on outdated material or present uncertain conclusions with a level of confidence that is not justified.

In financial contexts, these limitations have economic consequences.

An incorrect explanation of tax treatment, a fabricated financial reference, an inaccurate calculation or a misleading interpretation of investment performance can influence decisions involving real money.

The OECD’s 2026 policy paper, Artificial Intelligence and Personal Finance, identifies several opportunities and risks associated with the use of AI in personal finance. Its analysis includes accessibility, personalisation and decision support as potential benefits, alongside bias, hallucinations, commercial influence, data privacy and exclusion as areas of concern. It also emphasises that the long-term effects on financial wellbeing remain uncertain.[4]

These risks are not unique to AI. Financial misinformation, conflicts of interest and unsuitable recommendations existed long before generative AI. What changes is the speed, scale and presentation of information, as well as the possibility that users may not recognise the limitations of the system producing it.

The Spanish securities regulator, the Comisión Nacional del Mercado de Valores (CNMV), provides a particularly relevant example.

In its June 2026 Bulletin, researchers Ricardo Crisóstomo and Diana Mykhalyuk examined the ability of four large language models to generate stock-market forecasts. Their analysis identified recurring failures involving financial interpretation, calculations and the use of outdated, inconsistent or fabricated information.

The research also found that structured prompts, human oversight and the use of official regulated information could improve the quality of model outputs. However, the authors concluded that performance was not automatic or robust when models operated autonomously.[5]

This is an important distinction. AI may be useful in financial analysis, but its output still needs to be examined against reliable information and appropriate human judgement.

There is also a question of data. Financial questions can reveal income, debt, savings, investments, family circumstances and future plans. Sharing this information with a digital service requires an understanding of how the provider handles, retains and uses personal data.

Convenience should not remove the need for informed consent and privacy awareness.

5. The human factor: why the way we ask matters

AI-generated financial guidance is shaped not only by the model but also by the information and questions supplied by the user.

Consider two people asking for help with investing.

One asks: “What should I invest in to make the most money?”

The other explains their investment horizon, existing savings, debt obligations, tolerance for losses, liquidity needs and financial objectives, and asks for a comparison of different approaches.

The questions are different. The information available to the system is different. The responses may therefore be different.

A 2026 working paper by Taha Choukhmane, Tim de Silva, Weidong Lin and Matthew Akuzawa examines this relationship between the advice generated by AI systems and the questions people ask.

The researchers surveyed a demographically representative sample of 1,000 US adults and used the resulting prompts to examine advice generated by GPT-5.2 and Gemini 3 Flash. After excluding responses that did not meet their research criteria, the final survey sample contained 952 respondents. The researchers then simulated the lifetime financial implications of following the generated advice.[6]

Their findings are particularly interesting.

Overall, the AI-generated recommendations moved most simulated individuals closer to the broad principles of life-cycle financial planning, including participation in diversified equity funds, savings buffers and adjustments to equity exposure with age.

However, the recommendations also relied on relatively simple savings and withdrawal rules and did not always respond adequately to income or market shocks.

The researchers found that recommendations varied systematically according to financial literacy, previous AI experience and gender.

In their simulations, average wealth at retirement was around 5% higher when using prompts written by men, individuals with higher financial literacy or people with previous experience using AI for financial advice.

The researchers also examined the source of the gender differences in equity allocation recommendations. They found that approximately two-thirds came from differences in the prompts written by men and women, while the remaining third came from the model responding differently to the same question when gender was stated.

These are simulated outcomes, not observed differences in actual retirement accounts. The findings also relate to the particular models, prompts and assumptions examined in the research.

Nevertheless, they raise an important behavioural question: what happens when people’s existing knowledge, confidence and assumptions influence the quality of the questions they ask?

Financial decisions are already shaped by behavioural factors such as present bias, loss aversion, overconfidence, inertia and the tendency to rely on familiar information.

AI does not make these factors disappear.

Someone who is overconfident may accept an AI-generated investment explanation too readily. Someone who feels overwhelmed may delegate too much judgement to a system. Someone anxious about financial markets may repeatedly seek reassurance rather than examine the underlying risks.

There is also the possibility of automation bias: the tendency to place excessive trust in the output of an automated system, particularly when it appears technically sophisticated.

The challenge is not to eliminate human judgement from financial decisions. It is to strengthen it.

6. Women, Wealth and AI: financial agency in an AI-driven economy

The relationship between women, financial knowledge and AI deserves particular attention.

Women do not form a uniform group of financial consumers. Their financial circumstances differ according to income, age, education, occupation, family responsibilities, country of residence, access to financial services and accumulated assets.

These differences matter when considering who can benefit from AI and who may encounter barriers.

A 2026 study by Tae-Young Pak, published in the Journal of Behavioral and Experimental Finance, examined how 2,170 South Korean adults aged 25–59 used large language models for personal financial management.

The study found that 67.8% of respondents had used large language models for at least one of ten financial tasks. These included budgeting, savings, investment, tax filing, debt, insurance, housing, fraud detection, financial literacy and psychological support.

The study also found that usage was higher among men, younger adults, people with higher education and full-time workers. Differences by income, wealth and home ownership were not statistically significant.[7]

These results describe one South Korean sample and should not be treated as evidence of global gender differences in AI adoption.

Nevertheless, they raise questions worth examining. Are women accessing the same opportunities to develop AI-related financial skills? Do differences in time, confidence, digital access or financial education influence how these tools are used? Are AI systems providing equally useful explanations and recommendations across different financial circumstances?

The Choukhmane and colleagues working paper also identifies differences in AI-generated investment recommendations associated with gender and financial literacy. This suggests that the interaction between the person asking and the system responding deserves further investigation.

These questions were central to WEMATTER GLOBAL’s Women, Wealth & AI Conversation in Barcelona.

We brought together the subjects of financial agency, wealth creation and artificial intelligence because the relationship between them is too important to treat as a technology discussion alone.

There is the individual perspective: how can women use AI to understand financial concepts, explore investment opportunities and make better-informed decisions?

There is the economic perspective: how might AI influence women’s access to employment, entrepreneurship, income generation, investment and asset ownership?

And there is the institutional perspective: how are financial services, labour markets and economic systems changing, and whose interests are reflected in the technologies being developed?

Financial agency connects all three.

It is not simply the ability to use a digital tool or obtain an answer. It is the capacity to understand economic choices, evaluate alternatives, exercise judgement and participate in decisions that shape one’s financial future.

As AI becomes more integrated into financial services, the ability to question an algorithm, evaluate a recommendation and understand the implications of a financial decision becomes increasingly relevant to economic participation.

Financial empowerment in an AI-driven economy cannot be limited to access to technology. It must include the ability to understand, challenge and make informed use of it.

7. From AI assistance to informed judgement

Using AI responsibly in personal finance requires more than learning how to write effective prompts.

It requires a framework for evaluating the information received.

A practical approach is to consider five questions whenever AI is used to explore a financial decision.

1. What am I actually trying to decide?

Define the decision before asking for a recommendation. Are you trying to understand a financial concept, compare products, calculate a scenario or evaluate an investment?

2. What information is the answer based on?

Ask for the assumptions, data, sources and relevant dates. Distinguish between verified information, estimates and hypothetical illustrations.

3. What is missing from the analysis?

Consider whether the response accounts for your financial circumstances, obligations, time horizon, liquidity needs and exposure to risk.

4. What are the alternatives and trade-offs?

Ask what could change the outcome, what the disadvantages are and which assumptions matter most. A financial decision should not be evaluated solely on its most attractive possible outcome.

5. Who or what should verify the answer?

Check important information against original documents, official institutions, regulators and appropriately qualified professionals. For decisions involving tax, legal obligations, regulated financial products or significant investments, seek relevant professional advice where necessary.

The purpose of these questions is not to make every financial decision more complicated. It is to ensure that convenience does not replace understanding.

AI can help organise a decision. The individual still needs to understand what is being decided, why it matters and what the consequences may be.

8. The institutional question: who is responsible?

The growing use of AI in personal finance also raises questions that cannot be answered by individual users alone.

AI systems are developed and deployed by organisations. Financial institutions increasingly use automated technologies in customer service, credit assessment, fraud detection, investment services and other activities.

The design of these systems influences what consumers see, which options are presented and how decisions are explained.

This creates responsibilities for technology providers, financial institutions, regulators and policymakers.

The UK’s Financial Conduct Authority published The Mills Review in July 2026, examining how AI could reshape retail financial services for consumers, firms, markets and regulators.

The FCA identified four major AI-driven shifts: changes to firm operations, the evolution of consumer journeys, changes in competition and market power, and the amplification of fraud and cybersecurity risks.

The accompanying consumer research surveyed more than 5,000 UK retail financial-services consumers. One finding was that 20% said they would be likely to use AI capable of acting autonomously within predefined goals. The FCA translated this into an estimate of 11 million UK adults. This measures stated likelihood, not actual use.[8]

This points towards a further development: the transition from AI that answers questions to AI agents that may carry out tasks.

An AI assistant might explain a financial product. An agentic system could potentially compare products, initiate transactions or manage activities within permissions granted by the user.

The implications are different. When systems move from generating information to taking action, questions of authorisation, accountability, transparency, security and redress become even more important.

Financial institutions and technology providers must consider how consumers understand these systems and how responsibility is allocated when something goes wrong. Regulators must consider whether existing consumer-protection frameworks remain effective as these technologies evolve.

The individual should not carry the entire responsibility for identifying risks that are embedded in systems they cannot see or control.

Financial agency requires both individual capability and institutional accountability.

9. Technology should strengthen financial agency

Artificial intelligence is changing the way people encounter financial information and approach financial decisions. Its potential lies in making complex subjects more accessible, helping people explore alternatives and supporting a more informed relationship with money.

It may also make it easier to ask questions that people have previously avoided, reduce some of the cost barriers to initial financial learning and create opportunities for people to develop their understanding privately and at their own pace.

But access to an answer is not the same as understanding it. A personalised response is not necessarily a suitable recommendation. And a confident explanation is not proof of accuracy.

The challenge is to develop the knowledge, confidence and judgement to use these systems without surrendering the responsibility for financial decisions.

For women on the move, this question has particular relevance. Financial lives are shaped by changing countries, currencies, employment arrangements, family responsibilities, housing markets, pension systems and regulatory environments. The ability to interpret financial information across these circumstances can influence the choices available to them.

AI may become a useful participant in that process. It should not become the unquestioned authority.

At WEMATTER, we believe financial agency involves more than knowing how to earn, keep and grow money. It involves understanding the systems that influence financial choices and developing the capacity to make informed decisions within them.

The future of financial decision-making should not be about allowing technology to think for us. It should be about using technology to think more clearly, ask better questions and make decisions with greater understanding.

Because in an AI-driven economy, financial empowerment is not simply about having access to intelligence. It is about retaining the judgement to use it.


Sources and further reading

  1. Moorhouse, M., Goode, M., Cotte, J. and Widney, J. (2023). “Helping Those That Hide: Anticipated Stigmatization Drives Concealment and a Destructive Cycle of Debt.” Journal of Marketing Research, 60(6). https://doi.org/10.1177/00222437221146521
  2. StepChange Debt Charity (2026). “Shame and Stigma Keep People from Seeking Debt Advice, Research Finds.” 26 October 2026. https://debtcheck.stepchange.org/media-centre/press-releases/shame-and-stigma-debt-research.aspx
  3. Baeckström, Y. and Matkovskyy, R. (2026). “Financial Advice Behaviour: Humans versus AI.” Journal of Corporate Finance, 101, Article 103044. https://doi.org/10.1016/j.jcorpfin.2026.103044
  4. OECD (2026). Artificial Intelligence and Personal Finance. OECD Artificial Intelligence Papers, No. 62. OECD Publishing, Paris. https://doi.org/10.1787/2858fdf4-en
  5. Crisóstomo, R. and Mykhalyuk, D. (2026). “Large Language Models and Stock Investing: Is the Human Factor Required?” CNMV Bulletin, I/2026, pp. 59–84. CNMV Working Paper No. 96.
  6. Choukhmane, T., de Silva, T., Lin, W. and Akuzawa, M. (2026). AI Financial Advice: Supply, Demand, and Life Cycle Implications. Working Paper No. 4325, latest version dated 18 May 2026.
  7. Pak, T.-Y. (2026). “How Individuals Use Generative AI for Personal Financial Management.” Journal of Behavioral and Experimental Finance, 49, Article 101145. https://doi.org/10.1016/j.jbef.2026.101145
  8. Financial Conduct Authority (2026). The Mills Review: Review into the Long-Term Impact of AI on Retail Financial Services. Published July 2026. https://www.fca.org.uk/news/press-releases/fca-publishes-landmark-review-impact-ai-retail-financial-services
  9. Timmerman, I. (2026). “Learning From AI-Generated Financial Educational Content: Impacts on Subjective Knowledge and Financial Well-Being.” Journal of Consumer Affairs, 60(2), e70043. https://doi.org/10.1111/joca.70043

Leave a Reply

Your email address will not be published. Required fields are marked *

Share:

More Posts

The Opportunity Cost of Being a Woman

The opportunity cost of being a woman is staggering. Consider the hours spent on unpaid domestic responsibilities, caregiving, household management and the invisible work of keeping families and households functioning.

Send Us A Message