Back
The Impact of AI in Financial Decisions
29 August 2026 by Ed Freitas
When the Machine Is Fast, but the Decision Is Still Yours
Imagine an AI system that can approve a loan in seconds, detect fraud before a human analyst notices it, rebalance thousands of investment portfolios at once, and explain a quarterly variance before the finance team has finished its first coffee.
Now imagine that the same system confidently denies credit for the wrong reason, reacts to a deepfake, invents a regulatory requirement, or executes thousands of bad trades before anyone can press the stop button.
That is the uncomfortable reality behind the AI revolution in finance: the technology can operate at machine speed, but the consequences arrive at very human speed.
My upcoming book, The Impact of AI in Financial Decisions, is about that gap. It is not a coding manual for data scientists, and it is not another theoretical tour of AI buzzwords. It is a practical guide for financial leaders, managers, consultants, and practitioners who must decide where AI belongs, how it should be used, and when a human needs to step in.
Most importantly, it looks at those decisions through the lens of prompting. Not prompting as a party trick, but as a structured way to think, challenge assumptions, expose risks, and obtain useful work from generative AI.
From the AI Revolution to Real Financial Decisions
The book begins with the foundations: what AI actually means in a financial context, how it differs from traditional automation, and why financial institutions cannot treat it as just another IT implementation.
We will examine the complete life cycle of a financial AI project, from defining the problem and assessing the data to validating the model, deploying it, monitoring drift, and deciding whether a vendor's impressive demo deserves your money.
There is also a practical governance question running throughout the book: who is accountable when the model is wrong? Spoiler alert: it is not the algorithm.
Retail Banking, Credit, and Customer Experience
In retail banking, AI can compress decisions that once took weeks into minutes. It can support credit underwriting, detect fraud, personalize customer interactions, power next-best-action engines, and help service teams respond at scale.
But faster is not automatically fairer. The book explores black-box credit models, alternative data, proxy discrimination, explainable AI, adverse-action notices, chatbots, and the difficult balance between personalization and privacy.
We will look at what happens when a bank can make a decision quickly but cannot adequately explain it to the customer, or the regulator.
Investment Management and Trading
AI is also changing how investment professionals research companies, analyze earnings calls, interpret alternative data, construct portfolios, and react to market sentiment.
The opportunity is enormous, but so are the failure modes. An AI system can discover a signal that humans missed; it can also amplify herd behavior, overfit historical data, mistake online noise for insight, or respond to manipulated information before the truth catches up.
The investment section examines portfolio management, robo-advice, algorithmic trading, sentiment analysis, alternative data, and the role of AI as a devil's advocate against our own investment theses.
Risk, Compliance, and Financial Crime
Risk management is where the promise and danger of AI become especially clear.
The book covers fraud detection, anti-money-laundering analysis, regulatory compliance, model risk, cybersecurity, stress testing, and the growing threat from deepfakes and adversarial inputs.
AI can reduce mountains of false positives and help investigators find patterns that humans would never see. It can also introduce new systemic risks when many institutions rely on similar models, similar data, and similar automated reactions.
Corporate Finance and Strategy
The transformation is not limited to banks and markets. Inside corporate finance, AI is changing forecasting, financial planning and analysis, treasury, the continuous close, document processing, management reporting, and strategic decisions such as mergers and acquisitions.
We will explore where AI can remove repetitive work, where it can improve foresight, and where an authoritative-looking output can quietly conceal a bad assumption.
A beautifully formatted forecast is still wrong if the numbers underneath it are nonsense.
Ethics, Governance, and the Human in the Loop
A major part of the book deals with the questions organizations are often tempted to postpone: bias, fairness, explainability, privacy, accountability, transparency, and human oversight.
What should AI be allowed to decide? Which decisions require human approval? How do we test for unfair outcomes? How do we challenge a model that appears accurate overall but harms a particular group?
These are not abstract ethical debates. In finance, they affect who receives credit, whose transaction is blocked, whose portfolio is exposed, and who carries the cost when a system fails.
What Comes Next
The final part looks ahead to agentic AI, autonomous financial systems, embedded finance, quantum computing, new forms of fraud, and the skills financial professionals will need as routine analytical work becomes increasingly automated.
The book also includes a practical workbook filled with prompts, exercises, decision scenarios, and a capstone crisis involving an autonomous trading agent reacting to a deepfake announcement. The aim is not simply to tell you what AI can do. It is to help you think through what it should do, what could go wrong, and how you would respond.
This post is the starting point for a new series. In the articles that follow, I will take these topics one by one and turn them into practical discussions, examples, and prompts that finance professionals can apply to real work.
AI will undoubtedly make more of the analysis. It may recommend the action, draft the explanation, and even execute parts of the process. But the judgment, accountability, and values behind the final decision remain ours.
The machine is ready. The more important question is whether we are.