AI in Finance for Professionals Dealing with Errors, Overhype, and Constant Learning Pressure

Why It Matters

Understanding AI’s true capabilities and limits helps finance teams avoid costly errors, manage skill fatigue, and make strategic investments in data quality and training.

Key Takeaways

• AI accelerates data analysis but still requires human validation.
• Overhype leads to unrealistic expectations like instant three‑statement models.
• Data quality remains critical; AI can't fix garbage inputs.
• Continuous learning fatigue emerges from rapid AI feature updates.
• Training must balance AI tools with solid modeling fundamentals.

Summary

The Mod Squad episode brings together three finance‑modeling veterans to dissect the current state of AI in financial analysis. They explore how tools like Claude and emerging GPT models are being deployed for rapid report generation, survey synthesis, and even code‑first modeling, while emphasizing that the technology remains an augmentation rather than a replacement for human insight.

Key observations include dramatic time savings—hours instead of weeks for data‑heavy tasks—paired with persistent challenges: AI often produces plausible but inaccurate outputs, demanding rigorous double‑checking. The hosts debunk hype, citing examples such as a claim that a three‑statement model can be built in five minutes, which in reality took an hour for GPT‑5.2. A LinkedIn poll revealed that only 0‑20% of analysts currently rely on Claude, underscoring a gap between marketing hype and actual adoption.

Memorable remarks punctuate the discussion: “If your data is rubbish, your data is still going to be rubbish even if you throw a co‑pilot on it,” and “Excel isn’t dead; transitioning to cloud‑code models is a decade‑long journey.” These quotes illustrate the tension between enthusiasm for new capabilities and the practical limits of data quality and skill gaps.

The conversation concludes that firms must build robust data and knowledge foundations before scaling AI, and invest in continuous, realistic training. Overreliance on hype‑driven promises risks costly rework, while disciplined integration can unlock genuine efficiency gains for finance professionals.

Original Description

In this episode of The ModSquad, Paul Barnhurst, Ian Schnoor, and Giles Male share their real experiences using AI in financial modelling. They cut through the hype and discuss what AI actually does well, where it still struggles, and how professionals should be thinking about using it today. From building models to handling workflows, the conversation highlights both the value and the limitations of AI tools in real work.

Expect to Learn

Where AI actually helps in finance and modelling work

Why most “one-click solution” claims are unrealistic

The importance of checking and guiding AI outputs

How instructions and structure improve AI results

Here are a few quotes from the episode:

“It’s not a one-click solution. You still have to check everything.” – Giles Male

“You will have to understand every line to guide AI properly.” – Ian Schnoor

AI is powerful, but it’s not a shortcut to good work. It still needs guidance, structure, and strong fundamentals. The people who benefit most will be the ones who understand both the tools and the work behind them. For now, the best approach is simple: use it, test it, and don’t trust it blindly.

Follow Ian:

LinkedIn - https://www.linkedin.com/in/ianschnoor/

Follow Giles:

LinkedIn - https://www.linkedin.com/in/giles-male-30643b15/

In today’s episode:

[00:05] – Trailer

[02:06] – Current thoughts on AI after recent progress

[03:45] – Daily use of AI and time savings

[05:00] – Mental fatigue and keeping up with AI changes

[08:03] – Calling out AI hype and unrealistic claims

[12:30] – AI training challenges and business demand

[17:57] – “Eye of the storm” phase of AI development

[24:13] – Testing AI-built financial models

[30:52] – Why fixing AI models can take longer than building from scratch

[35:15] – Responsibility to challenge misleading AI claims

[38:41] – Using instructions to improve AI output

[42:59] – Final thoughts on AI, stress, and the future