AI & Professional Judgement

Deus ex Machina: AI & Financial Modelling

Sebastian Szameitat, AFM

Originally published 25 October 2024

Editorial note (29 October 2026): This essay was originally published in October 2024 and reflects the discussion and examples of that period. Its five original reference entries are reproduced as published; they describe sources but do not provide direct publication links. The original text has not been rewritten to reflect subsequent developments in AI.

Imagine a world where technology suddenly swoops in and solves your biggest problem – out of nowhere, a hero appears! In literature, this is called deus ex machina, a “god from the machine.” AI seems to be playing this role for financial modelling, where it promises to simplify complex tasks, save time, reduce risk, and improve accuracy. But can AI really deliver on these promises, or is it just an illusion?

AI is increasingly used in financial modelling, handling everything from mundane tasks like data input to sophisticated predictions. To decide if AI really is the “hero” it claims to be, let’s look at some different perspectives.

Automation: Time Saver or Skill Eroder?

First, let’s look at automation. Financial modelling is hard work – one wrong number in a spreadsheet can throw everything off. AI can automate tedious parts, like data entry and reconciliation, reducing errors and freeing analysts to focus on strategy. Companies like Deloitte and PwC note AI’s potential to save money by cutting down on manual work and avoiding costly mistakes.

However, automation has a downside. When AI takes over routine tasks, financial modellers might lose touch with essential skills and become overly reliant on technology. AI also needs precise instructions – “garbage in, garbage out” still applies! So while automation is valuable, it’s not a complete solution.

AI Predictions: Crystal Ball or Hazy Mirror?

One of AI’s most attractive features is its ability to predict outcomes. By studying patterns in historical data, AI can help financial modellers anticipate market changes. Major banks, like JPMorgan Chase, are already using AI to spot trends and even predict future movements in ways that seem magical.

But AI’s predictions have limitations. They’re only as good as the data they’re built on. When the COVID-19 pandemic hit, financial models struggled because they relied on historical data that didn’t account for such an unusual event. AI may excel at recognising patterns, but it can still be blind to unexpected changes. In unpredictable times, AI’s “crystal ball” can be misleading.

Risk Management: Guardian Angel or False Sense of Security?

Managing risk is one of the most complex parts of financial modelling, and AI seems like a “guardian angel,” especially with its real-time data tracking and anomaly detection. AI can quickly alert modellers to potential issues, helping to manage risk more effectively. Financial giants like HSBC and Goldman Sachs are already using AI for this purpose with impressive results.

But AI also has its limits in risk management. While it’s great at recognising trends, it might miss outliers – rare events that can be catastrophic. Another issue is responsibility. If an AI system recommends a risky decision, who is accountable if things go wrong? The human touch is still essential, even with AI’s guidance. AI may be a helpful tool, but it’s not infallible.

Simplifying Complex Tools: Helper or Complicator?

Financial modelling today uses many different tools, from Excel and Python to specialised platforms for forecasting. AI promises to unify this tool landscape, bridging gaps and automating data transfer between systems.

Imagine AI acting like an assistant who understands your goals, gathers the right data, and chooses the best tool for each task. Companies like Palantir are working on AI that can make this vision a reality, pulling data from different sources into one unified view.

Yet, this “helper” role can be challenging. Integrating all these tools can produce inconsistent results or incompatible outputs, and over time, AI’s involvement could make modellers lose touch with the details of each system. AI can simplify the workflow, but at a cost – it may unintentionally weaken the modeller’s connection to their own tools.

Ethical Considerations: The Role of Judgment and Transparency

AI introduces an ethical question: should it have so much influence in financial modelling? Many AI processes are “black boxes,” meaning that even experts don’t fully understand how certain decisions are made. In finance, where transparency is vital, this lack of clarity can be unsettling.

Financial modelling relies on human judgement, not just technical calculations. Clients, regulators, and stakeholders trust modellers to use ethical, balanced judgement – qualities AI doesn’t possess. AI can’t make moral decisions, nor can it consider the societal impact of financial choices. Until AI reaches a level where it can respect ethical and moral boundaries, it’s best viewed as a tool, not a replacement for human judgement.

Conclusion: Is AI Financial Modelling’s Hero?

So, is AI the deus ex machina of financial modelling? Yes and no. AI offers amazing potential to streamline tasks, improve accuracy, and speed up processes. But it’s still a tool, and like any tool, it’s only as good as the person using it. The best approach is to view AI as a powerful assistant, not a saviour. With the right balance, AI can help modellers work more effectively without taking over completely.

In the end, we should welcome AI into financial modelling – but cautiously. Financial modellers must stay alert, stay skilled, and never lose sight of the ethical and human aspects of their work. AI might be our industry’s most useful tool, but it’s not our hero. Not yet.

References

Deloitte Insights, “The Future of Financial Services: AI in Modelling and Forecasting”
PwC AI Case Study Reports
HSBC 2023 Annual Report on AI and Risk Management
Goldman Sachs, AI in Financial Forecasting 2024
Palantir White Papers on Data Integration for Financial Modelling