Blaming AI: The Dangerous Trend of Treating Technology Like a Person
When headlines read "AI fired employees," "AI rejected applicants," or "AI made the layoff decisions," they create a distorted narrative.
I don’t know why or when it started but we've begun talking about AI tools as though it's an independent actor, one capable of making decisions, exercising judgment, and bearing responsibility for the outcomes of those decisions.
It isn't. AI is software. An assistive technology created and configured to be deployed, monitored, and governed by humans.
And while AI systems can perform increasingly sophisticated and autonomous tasks, autonomy is not the same thing as accountability.
That distinction matters, because when we start treating technology as though it is responsible for human decisions, we risk making the humans who designed, deployed, approved, and relied on that technology less visible and that’s where the problem begins.
Let’s take the recent Meta lawsuit.
According to the complaint, an AI system allegedly evaluated employees primarily on productivity metrics, resulting in workers on protected leave being disproportionately selected for termination. News headlines quickly framed the story as one in which "AI made the decisions."
The headlines:
“Current and former employees sue Meta, alleging discrimination in using AI to conduct layoffs - workers allege that Meta’s “constellation of internal artificial-intelligence systems” failed to take approved absences into account when determining which employees to cut.”
“Meta’s AI-based layoffs allegedly targeted workers who had taken protected leave”
Public commentary focused almost entirely on "AI laying off humans." Many viewed it as another example of machines taking human roles.
Even if every allegation in the complaint were true, the central issue would not be that an algorithm behaved badly. The issue would be that people designed, approved, implemented, relied upon, and failed to audit the system.
Whether the AI generated recommendations or even ranked employees, humans remained responsible for:
selecting the variables
deciding which data to include
determining how recommendations would be used
validating the outputs
reviewing for legal compliance
Making or approving the final employment decisions
Another case in the headlines, Mobley v. Workday, alleges that Workday’s AI-powered applicant screening tools discriminate against job seekers based on race, age, and disability.
The arguments?
The company argues its tool does not make hiring choices and that enterprise customers retain total human control over the hiring process.
The customer’s argument "it’s the vendor's software".
The vendors are selling it as a decision tool and the customer is using it as such. Neither understanding or auditing the tool that’s been released with big promises until something goes wrong and reality sets in, the tool isn’t qualified to make decisions. The people who are qualified to make them are seemingly giving away that responsibility.
The same principle applies to every HR technology we've ever used. No one says an applicant tracking system hired someone.
No one says Excel or Google Sheets mismanaged the budget.
No one blames Gmail when a poorly written communication goes out.
The technology didn't decide. People used the technology poorly.
It’s time to strip away the sci-fi framing and restore accountability to where it belongs: with the people behind the prompts.
The Meta lawsuit is about human governance, not AI. The court has ruled that software vendors like Workday can be covered under anti-discrimination laws as employer agents if their tools manage core hiring filters.
Anthropomorphism Makes Accountability Disappear
Psychologists have long documented anthropomorphism, our tendency to assign human qualities, intentions, or emotions to non-human objects and systems.
Today's AI makes that tendency even stronger. We give systems names, assign genders, say "Claude thinks or decided." The companies are branding them as our co-worker. People routinely thank and apologize to these tools.
In order to increase adoption and user engagement some AI tools are intentionally designed to sound warm, empathetic, humorous, or conversational
As the tools influence trust and decision-making, we become more willing to describe it as though it possesses agency. Once that happens, accountability quietly shifts away.
AI has no agency. AI has:
no legal responsibility
no ethical judgment
no lived experience
no concern for fairness
no understanding of discrimination
no empathy
no fear of consequences
It predicts outputs from patterns in data. The moment we begin speaking as though AI independently "wanted," "decided," or "chose," we've abandoned reality in favor of a convenient narrative and what we need to remember:
🚫 The AI isn't paying damages.
🚫 The algorithm isn't appearing in court.
✅ The organization is.
Regulators have repeatedly emphasized that employers remain accountable for employment decisions assisted by AI, regardless of whether those decisions involve automated systems.
Ethical Responsibility Cannot be Outsourced
Beyond compliance lies a larger ethical question. Organizations increasingly describe AI as objective or unbiased. Yet every AI system reflects human choices:
which data to collect
what success looks like
which outcomes to optimize
which variables matter
which tradeoffs are acceptable
how much human review is required
Those are human integrated decisions. Not technical ones.
Language Shapes Accountability
Words matter. Instead of saying:
“AI fired employees.”
Say:
‘Decision makers relied on AI-generated recommendations without adequate human oversight.”
_______
"Can AI make this decision?"
Instead ask:
"What level of human accountability should always remain in this process?"
Take Action
Design governance before deployment
Document decision ownership
Conduct regular bias audits
Validate outcomes against protected groups
Require meaningful human review before consequential employment actions
Clearly identify who is accountable for every AI-assisted decision
The goal shouldn’t be to remove humans from the process. It's to make human responsibility unmistakable.
"Human Accountability Test" for any AI-assisted decision. Make sure you know:
Who configured the system?
Who approved its use?
Who validated the outputs?
Who had authority to override the recommendation?
Who is legally accountable for the final decision?