I get asked, often enough that I've started keeping a mental script ready, whether my research is "basically AI ethics." I understand why people ask, and I've stopped being annoyed by the question over time — it sits near enough to that field that the confusion is genuinely reasonable, not a sign anyone's misunderstood what I'm doing, just a sign the two fields haven't yet drawn a clean enough line between themselves for the average reader to spot. My honest answer is no, and I've come to think the distance between the two is exactly what makes my version usable by an actual founder this quarter, rather than eventually.
What the big question is actually built to answer
AI ethics, as a field, is doing serious and necessary work at a genuinely large scale: is a given AI system fair across different groups of people, is it transparent about how it reaches decisions, is someone accountable when it causes harm, is it robust and safe under conditions nobody fully tested for. The major frameworks reflect that scale. The OECD's AI Principles, adopted in 2019 and since referenced by dozens of national AI strategies, ask member governments to commit to inclusive growth, human-centered values, transparency, robustness, and accountability across entire AI ecosystems. The European Union's AI Act sorts systems into risk tiers — from practices banned outright to high-risk systems facing real regulatory obligations around documentation, oversight, and testing — because at the scale of an entire economy, you need a classification system before you can even start writing enforceable rules. This is the right altitude for that question. It has to operate at the level of societies, regulators, and entire categories of technology, because that's the level at which "is AI fair" actually needs answering.
It is also, almost by design, not a question most founders can act on this week. "Is our AI system fair, transparent, and accountable at a societal level" is a genuinely important question with a genuinely uncomfortable property for a small business: it's mostly answered by policymakers, auditors, and academic committees, on a timeline measured in years, not by the twelve-person team that just turned on a customer service chatbot and wants to know, today, whether that was a reasonable thing to do.
The narrower question I actually built my research around
My question sits underneath that one, not competing with it: specifically, what happens to a brand's identity and the trust it's built once AI starts producing its customer-facing communications at scale, and how do you govern that gap well. Notice what's missing from that sentence compared to the big-question version. I'm not asking whether AI in general is fair to society. I'm asking whether this specific brand's specific automated output still matches what that brand claims to stand for — a question small enough that a founder, not a regulator, can actually check it, and can check it today, not once a policy framework eventually catches up to the technology.
That narrowness isn't a limitation I'm apologizing for. It's the entire design choice. Big-question AI ethics has to wait for regulation, industry standards, and slow institutional consensus before most of its insights become actionable for an individual business. My narrower question was built specifically to skip that wait, because Brand Drift doesn't wait for regulation either — it's already happening inside businesses that think they're being efficient, right now, this quarter, regardless of what any eventual AI Act eventually classifies it as.
Where the two questions actually meet
They're not unrelated, and I don't want to overstate the separation. Sports Illustrated's fabricated-byline scandal — where the outlet published product reviews under the names of AI-generated authors who didn't exist, quietly swapped for new fake names once journalists asked questions, until the publisher's CEO and several executives lost their jobs over it — touches both registers at once. At the big-question level, it's a case study in AI transparency and accountability failure, exactly the kind of thing AI ethics scholarship is built to analyze, and exactly the kind of case that would eventually inform a risk-tier classification somewhere. At my narrower level, it's a case study in a specific brand's identity — decades of editorial trust — eroding through a specific, traceable governance gap in a specific vendor relationship. Both readings are correct. They're just operating at different altitudes, answering different people's different, equally real questions.
I think the field benefits from having both altitudes staffed properly, and I'd be suspicious of anyone claiming their altitude was the only one that mattered. What I'm not willing to do is let my research drift up into the big-question altitude just because it sounds more prestigious, because the businesses I actually talk to — wellness founders in Dubai, EdTech founders in Riyadh, consultancies in Casablanca — don't have a policy department waiting to translate societal-scale AI ethics into something they can check on a Tuesday. They need the narrower question, answered specifically, now.
Why I think this is the more useful question, not just the more convenient one
"Is this narrower or is this just smaller" is a fair challenge, and I've sat with it. My honest answer is that narrower, done properly, produces something big-question ethics structurally can't: a check an individual business can run themselves, this week, without waiting for a regulator, an industry standard, or an academic consensus to arrive and tell them what to do. Fairness at a societal scale is somebody else's job to build the machinery for, and I respect that machinery, even when it moves slower than any individual business would like. Whether your own brand's chatbot still sounds like you is nobody's job but yours, and it turns out almost nobody's actually been asking it, precisely because it fell into the gap between two fields that each assumed the other one was covering it.
Why founders keep mistaking one question for the other
I think part of the confusion I run into constantly comes from how AI ethics gets talked about in mainstream business media, where the language of fairness, bias, and transparency gets applied indiscriminately to everything from a hiring algorithm to a marketing chatbot, without much distinction drawn between systems making consequential decisions about people's lives and systems producing customer-facing text. A hiring algorithm that systematically disadvantages a demographic group is squarely a big-question fairness problem, and it deserves every bit of scrutiny the field applies to it. A chatbot that's drifted away from your brand's actual values isn't obviously a fairness problem in that same sense — nobody's being discriminated against by a slightly generic customer service reply — but it's still a real problem, just a different one, sitting at a different altitude, and answered by a different set of tools. Collapsing both into "AI ethics" as if they need the same remedy is, I think, part of why so many founders either over-worry about the wrong thing, treating a customer service tone problem like a discrimination lawsuit waiting to happen, or, more commonly, assume the whole territory is somebody else's job entirely and stop paying attention to any of it, including the much smaller part that was always theirs to own.
The part I still want AI ethics scholarship to hand me
I don't want this to read as a field guarding its own territory, because that's not actually how research is supposed to work, and it's not how I want mine to. The big-question work on fairness, bias, and accountability is exactly what eventually tells me whether the tools businesses are adopting were built responsibly in the first place, which matters enormously to whether governing their output well is even sufficient. My narrower question assumes the tool itself is reasonably sound and asks whether the brand deploying it is watching closely enough. If the tool isn't sound to begin with, no amount of brand-level governance fixes that, and that's precisely the gap the bigger question is supposed to close. I need that work to keep happening, and I'd genuinely rather see more of it funded and taken seriously, not less. I'm just not going to pretend my own research is doing it, when what it's actually doing is answering the much smaller, much more immediately checkable question sitting one level below it — the question a founder can act on today, with the tools they already have, rather than the question that requires an entire regulatory apparatus to answer properly on their behalf.
If "is this narrower question" describes exactly the gap you've been circling without a way to check it yourself, the free Brand Bloom Audit is built to answer it in fifteen questions. Take the audit →
References
European Union. (2024). Regulation (EU) 2024/1689 (Artificial Intelligence Act). Official Journal of the European Union.
Maiberg, E. (2023, November 27). Sports Illustrated published articles by fake, AI-generated writers. Futurism.
OECD. (2019). Recommendation of the Council on Artificial Intelligence. Organisation for Economic Co-operation and Development.