The Human Side of AI - part 3
Human Review Is More Than Quality Control
By: Kara Maddox
Communication doesn't only contain information. The effort behind it can become information too.
Imagine receiving two beautifully written messages from a company.
They're identical.
Then you're told:
Message A was created by a person.
Message B was generated using AI.
Nothing about the words changed.
But something about the story behind those words did.
And research increasingly suggests that people notice.
Effort is a signal
Across eight studies spanning online, laboratory, and field settings, Patil and Rice investigated what happens when brands disclose AI use in digital advertising.
Their central finding is striking:
AI-use disclosure reduced perceived effort.
And lower perceived effort subsequently reduced consumer engagement.
Critically, the advertising itself could remain constant. Consumers were responding to the production cue, not simply to differences in content (Patil & Rice, 2026).
That means customers can ask, consciously or not:
How much work did somebody put into this?
And the answer can affect how they respond.
AI disrupts an old signal
Why should effort matter?
Because humans have long used effort as information.
If someone spends considerable time preparing something, we may infer:
it mattered
they cared about the outcome
they considered it worth doing well.
Generative AI complicates that signal.
Keenan and colleagues describe exactly this problem in research communication. Traditionally, producing accessible, polished communication requires effort. AI can dramatically reduce that cost—but in doing so, it makes the amount of human effort behind a polished output harder for the recipient to infer (Keenan et al., 2026).
That's not necessarily bad.
Reducing unnecessary effort is one of AI's greatest benefits.
But it creates an interesting question:
What happens when the effort AI removes was itself carrying meaning?
The problem isn't efficiency
Businesses should absolutely use AI to eliminate labor that doesn't create customer value.
Nobody needs an employee manually typing:
Your order shipped today.
five hundred times.
But effort becomes more interesting when the communication itself is supposed to demonstrate investment.
An apology.
A recommendation.
A thoughtful response to a complicated complaint.
A message after a customer explains something personally consequential.
In those cases, the customer may care not only about the final words.
They may care that someone thought before sending them.
Apologies make the signal visible
Glikson and Asscher examined AI-mediated workplace apologies and found that extensive AI involvement reduced perceived authenticity compared with no AI involvement. Limited forms of AI assistance did not produce the same authenticity penalty (Glikson & Asscher, 2023).
Their theoretical discussion is particularly useful because it treats effort as part of the signal behind authentic communication.
An apology that costs someone something—time, thought, vulnerability, attention—may communicate differently from one produced with almost no human investment.
Again, that's not an argument for making people work harder for the sake of it.
It's an argument for distinguishing:
unnecessary labor
from
meaningful investment.
Human review could change the signal
This is where the research reaches an interesting edge.
Suppose a customer sees:
This message was generated using AI.
Now compare:
This message was generated using AI and then reviewed and revised by a member of our team.
Both disclose AI involvement.
But the second adds another piece of information:
A person did something afterward.
That could matter for several reasons.
Maybe human review signals:
effort
quality control
attention
judgment
or
responsibility.
We don't yet know.
And that's precisely the point.
The research establishes that AI use can alter perceived effort.
It does not yet adequately establish whether meaningful downstream human involvement restores that effort signal.
That's a question worth testing.
Don't add human work everywhere
The takeaway here is not:
Humans should rewrite everything AI produces.
That would miss the entire value proposition of AI.
Instead, ask:
Does human effort carry meaning in this interaction?
For a shipping notification?
Probably not.
For a serious complaint?
Maybe.
For advice?
Possibly.
For an apology?
Research suggests we should at least ask.
Because sometimes efficiency removes labor nobody valued.
And sometimes it may remove evidence that somebody showed up.
There's a difference.
Research note
This article is Part 3 of KJMdigital's The Human Side of AI series.
Research supports the proposition that AI disclosure can reduce perceived human effort and that effort can function as information about the communicator.
The idea that disclosed downstream human review restores that signal remains a hypothesis—not an established finding.
That's where the next research question begins.
References
Glikson, E., & Asscher, O. (2023). AI-mediated apology in a multilingual work context: Implications for perceived authenticity and willingness to forgive. Computers in Human Behavior.https://doi.org/10.1016/j.chb.2023.107922
Keenan, M., Karachiwalla, N., Koo, J., Mwangi, C., Breisinger, C., & Kim, M. (2026). Man vs. machine: Multi-country experimental evidence on the quality and perceptions of AI-generated research blog content. PLOS ONE, 21(3), e0342852.https://doi.org/10.1371/journal.pone.0342852
Patil, R. K., & Rice, D. H. (2026). Is that your best effort? The impact of AI-use disclosure on consumer engagement. Journal of Business Research, 215, 116354.https://doi.org/10.1016/j.jbusres.2026.116354
Comments