The Boardroom AI Identity Crisis: Not All AI is the Same AI
Has AI officially arrived in Wealth and Asset Management?
I mean, it’s everywhere. It’s in mission statements, roadmaps, boardroom discussions, a prerequisite for many jobs, and increasingly, becoming seen as headcount. But while the appetite for AI is high, the understanding of what AI actually is and isn’t, is often muddled.
Specifically…Traditional AI (used to streamline, automate and bring structure to processes) vs Generative AI (which imitates human reasoning and generates new insights).
These are not interchangeable tools. They serve different purposes, require different inputs, and deliver different outcomes. And so for leaders and people trying to drive change across business and functions i.e. Market Data, Performance, Risk, Research, or Operations, distinguishing between the two is essential.
Traditional AI: The Automation Engine
Standardly referred to as ‘Predictive’ or ‘Analytical AI’, this is the workhorse and primarily focuses on automation, doing its best work in structured environments where rules are set and repeatable. These systems don’t “think” or create; they execute logic at scale, faster and if the right parameters are set, then more accurately than people.
Example use cases would be:-
Market Data: Analyse structured spend data by vendor, track renewal dates, monitor license utilisation vs entitlement
Operations: Automated trade matching, reconciliation exception handling, post-trade reporting validation.
Performance: Return calculation engines, benchmark comparisons, attribution modelling.
Risk: Real-time compliance alerts, rule-based exposure monitoring, VaR calculations.
Research: Screening financial instruments based on filters (e.g. P/E ratio, yield thresholds).
In each of these, the AI is pretty simple: feed it data and it will give you the same result every time based on the rules given.
Generative AI: The Creative Collaborator
Built on large language models (LLMs) and neural networks, they copy human language and create content based on patterns it has learned, typically from large datasets.This isn’t about automation; it’s about enhancement. It helps people explore ideas and navigate complexity.
Example use cases would be:-
Market Data: Summarising vendor spend patterns, comparing vendor contracts side-by-side, drafting RFP questions
Operations: Drafting client comms or internal incident summaries based on raw input data.
Performance: Generating initial fund commentary drafts for factsheets or reports.
Risk: Creating narrative summaries of stress scenarios or risk committee outputs.
Research: Summarising market outlooks, generating views on economic trends, or translating complex data into plain English.
It’s powerful but not foolproof, as the sources and output require constant validation.
For a deeper dive into TradAI vs GenAI, I’ve put together my own list of use cases and outputs from my most recent consultancy project with PwC and previous roles within Market Data here.
Why This Distinction Matters
Too often, senior leaders look at Generative AI and expect it to deliver automation. Or they see Traditional AI and assume it can provide insights. But this can lead to, among other things:-
Inappropriate application and use of AI in areas requiring high accuracy but the tech used focuses on association (or probability)
Misallocated resources where companies invest in for example GenAI for tasks that suited to traditional automated tasks
Risk of reputational or regulatory errors, if the tech is misapplied or misrepresented
Data leaks and privacy violations where sensitive data is uploaded into public tools, resulting in data breaches
Poor adoption rates and disappointed users in the business
Add to this, the introduction of The Digital Operational Resilience Act (DORA) in January 2025, where now, the landscape of what it means to be operationally resilient has changed.
Regulators now expect firms to have a granular understanding of their technology dependencies. For example: if a firm relies on GenAI for key traditional tasks i.e. compliance monitoring or contract auditing, without acknowledging the risks, this isn’t just a tech error, but also an operational resilience gap. DORA is about predictability and reliability. Traditional AI is logic-based and repeatable; therefore, through process documentation, it will meet regulatory standards. Whereas Generative AI is creative in nature, so it’s a risk that needs to understood and managed.
Having a clear understanding of the difference in AI and having the right governance, not only means understanding the regulatory impact behind the technology, but you’re also more likely to use the right tool for the right task with the correct provisions empowering teams and people to do the same.
The Human Side of AI: Two Critical Questions People Must Ask
For all the headlines about AI, the reality is that it was designed by and is for the benefit of people. And if you’re not using it then you’re missing out. But with AI increasingly being used across businesses to improve operational efficiency, client experience and decision making, we must consider the risks and challenges that come with AI adoption and usage.
1. What does AI mean for our mental capacity and cognitive thinking?
One risk of relying too heavily on AI, especially Generative AI, is the gradual erosion of our own analytical and critical thinking skills. Using AI is very much encouraged, but if people simply accept outputs without scrutiny, they risk outsourcing not just the tasks but also judgement.
This is referenced across a range of studies in Nature Human Behaviour and publications by MIT Sloan discussing in various ways, the cognitive impact, where over-reliance can impact the brain’s own problem solving effort because it expects technology to do the heavy lifting.
The danger? Less questioning, less creativity, and a decline in independent decision-making over time.
However, the flipside of that are the tasks that AI is least likely to replace, because of the need for human qualities and the reliance on empathy, judgement and ethics. When it comes to generative ideas or concepts, AI also struggles with subjective beliefs where the data does not give it a definitive answer. This is where people are highly effective, because we can make decisions not because the data tells us to, but because we have the judgement to understand what the decisions is and why it needs to made.
For leaders, this means AI adoption should go hand-in-hand with embedding a healthy challenge culture: encouraging teams to interrogate outputs, compare against their own reasoning, and treat AI as a starting point and not the final word.
2. Is the future of AI about asking the right questions?
The real power of AI lies not just in the answers it gives, but in the questions we ask it. Framing problems effectively, providing clear inputs, and probing outputs critically, is what makes AI useful.
In my own work within Market Data, I’ve used AI directly for:-
Summarising Contracts: Pulling out key clauses, but only if I know what contractual nuances matter from the licensing agreements.
Auditing Licensing: Highlighting exceptions, but it’s my job to define what “compliance” should look like in the first place.
Forecasting Fees: Modelling different fee scenarios, but it’s my role to sense-check assumptions against market benchmarking.
Each of these requires not just technical literacy, but also the ability to probe and to ask: What’s missing? Where could this be wrong? How do I validate this?
Final Reflection
As AI becomes more embedded into businesses and processes, the real differentiator won’t be who has access to the tools, but who uses and how to use them with judgment, curiosity, and detail.
Traditional AI and Generative AI both bring huge value, but only when leaders and businesses frame them correctly and ensure their teams and people stay engaged, raising the bar for critical thinking and questioning, to elevate the art of what is possible.
Note: TradAI is not a standard industry term. It is commonly referred to as “Predictive AI” or “Analytical AI”. However, for ease and comparison within this blog, I have referred to it as TradAI
