Bad news for anyone comforting themselves with the idea that people want humans doing search marketing: they mostly don’t care.
Harvard Assistant Professor James Riley asked 2,357 Americans to rate how morally objectionable it would be to hand 940 different occupations to a machine, on a 1-to-7 scale. Search marketing strategists came in at 2.31. Of the 10 occupations Harvard charted, only file clerks scored lower.
For comparison: clergy scored 5.91. Childcare workers scored 5.86.
What the numbers actually say
The research is bundled into “AI in 2026: From Adoption to Agentic,” a February 2026 roundup from HBS Working Knowledge that collects five previously published pieces. Riley’s occupation study, published October 2025, is the headline act.
The 2.31 score is only half the story. Riley found the public currently supports fully automating roughly 30% of the occupations tested, based on what AI can do today. When respondents were asked to imagine a more advanced AI that beats humans at lower cost, support jumped to 58%.
Only about 12% of occupations drew strong moral resistance no matter how good the machine got – clergy, childcare workers, athletes. Another 42% left people ambivalent.
Riley’s own reading: resistance to automation is mostly about whether the tech is capable yet, not about principle.
That’s a much shakier defence than “people believe humans should do this work.”
Preference follows perceived competence
A second study in the roundup makes the same point from a different angle. Assistant Professor Elisabeth Paulson and UC Berkeley’s Kirk Bansak ran a conjoint experiment with 9,000 participants, asking them to pick a human or an algorithm to approve a loan or decide a defendant’s pretrial release.
On average people leaned human – by 4.3 percentage points on loans, 7.6 points on pretrial release. But the split underneath is the interesting bit:
- Among people who already believed algorithms outperform humans, 56% picked the algorithm for pretrial release and 54% for the loan.
- Among people who believed humans were better, 63% and 59% went with the human.
- Fairness – equal treatment across racial groups – was the least important factor in how anyone judged either decision-maker.
Paulson’s take is that if you can demonstrate genuine accuracy gains without other metrics slipping, “that’s probably sufficient.”
Translation for marketers: the human preference isn’t a moral floor. It’s a scoreboard reading. Change the score, change the preference.
The competence gap is closing
Harvard’s Raffaella Sadun, Karim Lakhani and coauthors tracked 791 product developers at Procter & Gamble – some solo, some in teams, some with an internal GPT-4 tool, some without.
Ideas ranking in the top 10% for quality were three times more likely to come from AI-assisted teams than from unassisted individuals. Employees using AI also reported more enthusiasm and energy, and less anxiety and frustration, than those working alone.
Idea generation and content production is precisely the work search marketers bill for.
Meanwhile, a technical note from Tsedal Neeley and Expedia Group’s Ritcha Ranjan sketches agentic AI operating as chief of staff, competitive intelligence analyst and executive coach with minimal oversight once configured. Neeley advises leaders to start with “no-joy” work – the repetitive tasks nobody wants – before handing over anything higher stakes.
Sensible. Also a fair description of how automation creeps upward once it proves itself on the boring stuff.
What to do about it this quarter
1. Put a verifiable human name on anything AI touched. Not “Editorial Team.” A person with a LinkedIn profile, credentials and a track record a reader or a crawler can cross-check. Paulson’s data says preference tracks perceived competence, so give the byline something to attach to.
2. Publish your receipts, not your process. If your content or SEO program produced measurable outcomes, put the numbers in the piece. That’s the accuracy demonstration that shifts belief – and it works in your favour just as easily as it works against you.
3. Automate the no-joy layer only. Internal link audits, meta description drafts, log file triage. Keep a named human on anything touching reader trust or client money.
Why this matters
Much of our industry has assumed E-E-A-T and named human bylines get rewarded because the public has some residual stake in search staying human. Harvard’s own numbers say that stake is thin.
What’s protecting search marketing is a competence gap, not a conscience. Google’s systems – and increasingly the citation behaviour of AI answer engines – are running the same test Paulson’s respondents ran on loan officers: is the human version still demonstrably better?
Build your work so the answer stays yes.
Source: Search Engine Journal
