A featured contribution from Leadership Perspectives, a curated forum for banking, financial services, and fintech leaders, nominated by our subscribers and vetted by the Financial Services Review Editorial Board.



Taking Responsibility for AI
A client receives a beautifully written response in seconds. The tone is reassuring. The explanation sounds authoritative. The information is wrong.
Now reverse the situation.
A financial professional receives a meticulously worded request from a client. It includes technical terminology, confident conclusions and instructions to proceed. The client used AI to prepare it and may not understand exactly what they are asking anyone to do.
Two polished messages. Two opportunities for misplaced confidence. One relationship carrying the consequences.
In financial services, we need a more candid conversation about AI. Firms must take responsibility for the answers they provide. Clients must take responsibility for the information, questions and instructions they submit.
My standard applies in both directions: automate the task and own the consequence.
Fluency Is Not Understanding
Working across finance, operations and onboarding, I value tools that help people communicate clearly and navigate unfamiliar processes. AI can help someone organize scattered thoughts, prepare questions or understand terminology well enough to begin a productive conversation.
The danger arises when polished language conceals uncertainty.
A client may submit a request containing assumptions they have never verified. A professional may respond without recognizing that the wording overstates the client’s understanding or intent.
Suddenly, both parties are working from a foundation neither has checked.
If your AI writes the question, my AI writes the answer, and neither of us reads closely. We have created a very efficient meeting that neither of us attended.
Before acting, someone needs to establish what is actually being requested and why.
Keeping Human Judgment Central
Financial professionals cannot allow the confidence of an AI-generated answer to exceed the confidence justified by the underlying facts. We must verify material information, protect confidential data, respect the limits of our authority and involve qualified colleagues when appropriate. When an answer remains uncertain, we should say so plainly.
“A faster conversation is useful. A conversation both people actually understand is worth far more.”
We also need to recognize when a technically sophisticated request requires a basic clarifying conversation. ‘Can you tell me what you are trying to accomplish?’ may be the most valuable question we ask all day.
Professional judgment includes noticing when the requested action may not match the intended outcome. Excellent service requires enough attention to identify that gap before it becomes someone else’s problem.
Helping Clients Use AI Responsibly
Clients should never need financial expertise to ask for help. Explaining unfamiliar requirements is part of our responsibility.
But clients have a meaningful role in protecting the accuracy of the exchange. Read what AI prepares before sending it. Confirm the names, amounts, dates and circumstances. Remove claims you cannot support. If you do not understand a term, ask about it rather than presenting it as your instruction.
Tell us your objective, identify what you know and distinguish that from what a tool suggested.
‘AI suggested this approach. Does it fit my situation?’ opens a useful conversation. ‘This is what you are required to do’ creates a different problem when the statement rests on an unverified answer.
That distinction protects your interests. It gives the professional an opportunity to evaluate the premise before responding to the demand.
The same care applies to sensitive information. Convenience should not replace checking whether an AI tool is appropriate for the information being entered.
Shared Accountability Does Not Mean Equal Expertise
There is an important boundary here: asking clients to use AI responsibly does not relieve professionals of their obligations.
We bring specialized knowledge to the relationship. We remain responsible for applying it thoughtfully, explaining limitations and questioning inconsistencies.
Clients contribute accurate facts, candid explanations and confirmation of their intentions. Professionals contribute informed judgment, appropriate safeguards and clear guidance.
Both sides must be willing to say, ‘That is not what I meant,’ or ‘I need to check that.’
Those admissions prevent mistakes. We should make them easier, not embarrassing.
Trust Requires Someone to Pay Attention
Empathy means remembering that a confusing request may come from someone trying hard to navigate unfamiliar territory. Integrity means addressing the confusion honestly. Professionalism means ensuring the next step is understood and appropriately authorized.
AI can help both sides prepare for that exchange. It cannot take responsibility for the relationship. The leadership challenge is to build a culture where speed remains accountable to accuracy and polished communication remains accountable to truth.
Before pressing send, each of us should ask: Do I understand this? Is it accurate? Does it reflect what I intend?
A faster conversation is useful. A conversation both people actually understand is worth far more.