Don’t be a ‘meat proxy’ – why advisors shouldn’t send AI-produced content to clients

There’s a new insult making the rounds in tech circles, and financial advisors should pay attention.

A “meat proxy” is someone who takes what an AI tool produces and passes it along without reading it, understanding it or checking that it’s correct. The term was popularized this summer by German software engineer Niklas Gruhn.

A month later, Shopify chief executive Tobias Lütke complained that employees were lobbing unexamined AI-generated “slop grenades” at colleagues, leaving whoever catches them to do the checking.

We all know someone who mass-produces this stuff with no grasp of the topic and no effort to verify it. In most industries, that wastes time. In ours, it’s a real professional problem.

Certified financial planners and qualified associate financial planners are bound by FP Canada’s Standards of Professional Responsibility. Its Code of Ethics requires competence and diligence and its rules require CFPs and QAFPs to exercise reasonable and prudent professional judgment and to advise only where they’re competent in doing so.

Even when work is handed to others, the financial planner remains ultimately responsible and liable for it. That authority can’t be delegated to a person and it certainly can’t be delegated to software.

The FP Canada Standards Council made this very explicit in its February guidance on generative AI. AI should augment, not replace, professional judgment. Before entering client information, we should understand a platform’s privacy, security, retention and data-use practices. Outputs should be checked against authoritative sources before we rely on them. In other words, we need to understand what goes in, what comes out and what happens to the data in between.

The implication is simple. If you can’t verify or judge professionally whether an output is valid, you have no business using it. “The AI got it wrong” isn’t a defence; it’s an admission of guilt.

The same standards require reasonable and prudent supervision of any subordinate or third party, and the financial planner is ultimately responsible for the work they perform. If your associate or paraplanner is a meat proxy, so are you – you just don’t know it yet.

Every firm needs a written AI policy covering which tools are approved, what client data can go into them and what review happens before anything reaches a client. Then, it needs to enforce it.

Should this stop you from using AI? Absolutely not. As with any tool, the quality of the output depends largely on the craftsperson who wields it, and there are good and bad use cases.

I had an AI tool review dozens of my articles and build a style file – my tone, my preferences, my propensity for bad jokes – so its drafts sound like me. When I use AI to research and compose in my own voice, under my direction, with my review and edits, how is that different from working with a co-author? A chart of a client’s cash flow doesn’t need a disclaimer because software helped draw it. The professional is doing the thinking and owning the result.

Contrast that with an advisor who asks AI for a detailed tax or estate strategy they don’t quite understand, can’t tell whether the tool has made something up and then sends it to a client anyway. That isn’t validated work; under the standards, it’s unacceptable work. If you think blaming the AI won’t fly with regulators, imagine how it’ll fly in court.

There’s a second cost to being a meat proxy. Once people figure out you’re sending them material you never read, they stop reading it, too.

Consider Amazon.com Inc., famous for its six-page memos. Now that anyone can produce six pages in minutes, reading the memos has apparently become too much for some at the company where, rumour has it, some of the documents are being tagged “AI;DR” – AI, didn‘t read.

Expecting someone else to read something you didn’t bother to is a bridge too far. Worse yet, it undermines your professionalism and raises the question of why the client is paying you at all if software can give them similar unvetted answers.

If anything, in a world in which AI gives clients access to any answer they want, the value advisors bring to the table is understanding whether the output clients are showing you is accurate and fits their situation.

Regulators write rules for the worst behaviour they see. If enough advisors become meat proxies, the answer will be blanket requirements and compliance overhead that land on everyone, including those using these tools responsibly.

So, read what you send. Understand it. Check it. If you can’t vouch for it, don’t forward it. Here’s a simple test: Could you explain every recommendation to the client without the document in front of you? If not, it isn’t ready.

Don’t be a meat proxy. It wastes everyone’s time, costs you your credibility and puts you on the wrong side of your professional standards.

Jason Pereira is a senior partner and financial planner at Woodgate Financial Inc. in Toronto.

Editor’s note: Although it may be okay for advisors to produce material co-authored by AI, The Globe and Mail prohibits the use of generative AI for writing or editing. Read our generative AI guidelines for freelancers here.