I took a photo of the faucet in my upstairs bathroom, sent it to Gemini, and asked what would actually fit the space. No need for a Pinterest board or three tabs of Wayfair reviews. Just a picture and a question.

Gemini walked me through the whole configuration I needed (three-hole versus a single-hole with the two side handles) and talked through solid finishes. It helped me decide on something that matched what I actually wanted: a specific coat and finish, a specific placement, trade-offs I cared about.

I bought it. It arrived. And because I am, apparently, a walking Zillennial cliché, I asked my dad to install it.

He picked up the box before he even said hello. Turned it over. Asked if I’d measured. Asked if I’d checked the placement.

This wasn’t his first rodeo with my sink. The previous faucet was pricey, had great reviews when I bought it, and the finish started going copper and black before the two-year mark 😓

So when he looked skeptical, he had a real reason to be. Last time I trusted a five-star rating and a nice photo, and it cost me twice.

Then I told him AI picked it. Not “I picked it with some help,” but AI picked it.

He put the box down, said something like “okay, let’s install it and see,” and started working. No more questions about measurements. No more hedging.

The skepticism didn’t get resolved, but it just stopped mattering once “AI said so” was in the room.

What Actually Changed in That Conversation

Nothing about the faucet changed when I said that sentence. It was the same box, measurements, and finish I’d already described to him minutes earlier. The only thing that moved was who got credit for the decision.

Once it wasn’t just me anymore, my dad’s whole posture shifted from “prove it to me” to “let’s find out.”

That’s a strange thing to consider, especially because nothing about my side of the equation changed. The reasoning, measurements, and taste in finishes are all still mine. The only difference is that I ran it through Gemini first.

But somehow, once I could say the decision wasn’t only mine, it stopped needing defense.

This post is about that shift, and it’s bigger than one faucet.

AI isn’t just a tool people reach for anymore. It’s turning into something closer to a decision-maker, one whose word carries weight that a person’s own reasoning, stated plainly, increasingly doesn’t get on its own.

Why “The Algorithm Said So” Works On Us

There’s a name for what happened with my dad, and it’s not new. Researchers have been documenting it since before most people had touched a chatbot.

A 2019 study published in Organizational Behavior and Human Decision Processes found that people relied more on identical advice when they believed it came from an algorithm than when they believed it came from a person 😬

The researchers called it algorithm appreciation, and it held up across six separate experiments from visual estimates and song popularity forecasts to even predictions about romantic attraction.

Part of what’s going on is something researchers call the illusion of objectivity. You can push back on a person, ask them to justify their reasoning, point out their bad day, question their bias. That works because you know a person’s judgment can be swayed by mood, fatigue, or self-interest.

However, a confident answer from an algorithm doesn’t invite that same argument, because it doesn’t read as a judgment call at all. That response is less like an opinion and more like a fact you’d have to be pretty bold to challenge.

And that’s pretty much what happened on my bathroom floor. My dad wasn’t wrong to be skeptical. He’d watched a “great reviews” faucet fail in under two years. But once the decision had AI’s name attached, the confidence read as different in kind, not just in degree. It wasn’t “my daughter thinks this is right,” but more like “the math checked out.”

Tip 💁‍♀️
This is about how differently the same conclusion gets treated depending on who, or what, gets credited with reaching it. Not really about AI being smarter than the person it’s replacing in the conversation.

Who Gets the Credit, and Who Gets the Blame

If that faucet fails at the two-year mark, same as the last one, who’s responsible? Me, for making the purchase? Or Gemini, for helping me decide on it?

Research on AI-assisted loan decisions found that when people were approved, they credited the human evaluators. When they were rejected, they were quicker to blame the algorithm.

A study covered by the University of Illinois Gies College of Business revealed people wanted good outcomes attributed to themselves and bad outcomes pushed onto something external, and AI makes a convenient place to push them.

I know where I loudly laid the blame “just in case” it didn’t work, and it totally falls along these findings 😏

The researchers also noted people don’t extend AI the same grace they extend human error when AI makes mistakes. A person can learn from a mistake. An algorithm, in most people’s minds, apparently can’t.

That’s the quiet trade happening in my faucet story, too. If the new faucet works out, I get to say I made a smart call. If it doesn’t, “AI said so” stops sounding like backup and starts sounding like an excuse. Either way, credit flows to the human on success, blame flows to AI on failure.

Hey! If you’ve ever wondered why a coding assistant’s confident answer feels harder to question than a colleague’s, you’re running into the same pattern from a different angle.

Small Decisions, Outsourced By Default

Sadly, the faucet wasn’t a one-off. I’ve noticed myself reaching for AI to make decisions that used to happen without it. Things like what to wear to something. Which version of an email sounds better.

Whether a purchase is actually worth it or looks worth it in a five-star review. (Though for this, I give it to AI since it saves hours of browsing.)

I’m clearly not the only one. PartnerCentric’s 2026 consumer survey found that nearly 1 in 4 consumers are planning to make AI shopping their default way to shop this year. Most said it meaningfully reduces decision fatigue (which I stand by).

Gen Z has turned outfit-picking into a genre of its own. Fast Company reported that ChatGPT gets roughly a billion searches a week, with people asking it what to wear to dinner the same way they’d ask a friend.

None of that is inherently a problem. Decision fatigue is a real cost, and if AI cuts down the number of low-stakes choices eating your mental bandwidth, that’s a legitimate trade.

I don’t think the issue is that I asked Gemini about a faucet.

The issue is what happened after.

The answer needed AI’s name attached before it counted as settled, even with someone who already trusted my judgment on most other things.

Note 👀
If a tool just made the choosing faster, that would be one story. What actually happened is that the tool’s presence changed how solid the choice felt once it was made, to someone who wasn’t even in the room when it got made.

Explore: You Need To Work Smarter, Not Harder, With AI

Not Being Taken Seriously Without the AI Stamp

I’ve come to this uncomfortable realization in conversations that have nothing to do with home repair, when a fair point made in my own voice gets a shrug, while the same point with “I asked AI about this, and it agrees” gets a nod.

I don’t believe it’s that people think I’m wrong, but more so that my own reasoning, on its own, doesn’t seem to carry the weight it used to carry as a stand-alone argument.

Somewhere along the way, “here’s what I think” and “here’s what I think, backed by AI” stopped being the same claim in other people’s eyes, even though the thinking behind both is identical.

The research on algorithm appreciation actually predicts this.

If people already default to trusting algorithmic output over an equivalent human opinion (and I mean really trusting, not just saying), attaching that source to my opinion should change how seriously people take it, even when the opinion itself hasn’t moved an inch. Its content has not improved, but the perceived source did.

My dad didn’t respect my measurements more once AI was involved. He respected the label more than the reasoning it was labeling.

Let’s take a moment to think why that’s a strange position to be in.

I basically needed to borrow credibility from something that, in my faucet story, was working entirely off the preferences and details I gave it in the first place.

The “objective” source was repackaging my own judgment back to me, and somehow that repackaging is what made it count.

Related: This Is The Reason Why You Sound More Like AI

What I Might Be Trading Away

Researcher Michael Gerlich’s 2025 study of 666 participants, covered by the APA, found a negative correlation between frequent AI tool use and self-reported critical thinking, with something researchers call cognitive offloading as the likely mechanism.

The more reasoning tasks get handed to AI, the less practice a person gets doing that reasoning independently.

I recently shared a milder version of this during my PR rounds and the insecurity it’s breeding in many professionals’ lives.

One detail from that same body of research that particularly stuck with me is a study that found three months after AI support was introduced, clinicians’ ability to detect tumors without AI assistance had dropped measurably 😳

That’s not someone choosing convenience over a faucet. It’s a skill quietly eroding because it stopped getting exercised on its own.

I don’t think that means offloading is automatically bad. Later research on the same topic found the effect depends heavily on how people use the freed-up mental space, not just whether they use AI at all.

Tip 📍
We still need to be honest about the trade here. When the AI-backed version of a decision starts outperforming the version that’s just mine, I lose more than time saved. I lose practice trusting and defending my own thinking.

Explore: Am I Still a Developer If I Didn’t Write the Code? for more real-life evidence of cognitive offloading.

It’s a Wrap

The faucet isn’t installed wrong. Gemini didn’t lie to me, and my dad wasn’t unreasonable for doubting a purchase after watching the last one fail. There’s no “mistake” to grapple with here.

What I’m pointing out is smaller and stranger than any of that because it explores how the decision became more legitimate the moment it stopped being only mine.

Not more correct. More legitimate.

Those aren’t the same thing, and I don’t think I’d noticed the gap between them until I watched my dad’s whole posture change over one sentence.

What’s true is that AI helped me make a better call on that faucet. The reasoning was sound, and I’d make the same choice again.

But I’d like to get back to a place where my own reasoning, said in my own voice, doesn’t need a co-signer to be taken seriously.

Right now I’m not sure that place still exists for most decisions, small or otherwise. It’s not necessarily because I’ve made worse judgments; rather, the environment in which I make decisions has changed without my input.

I haven’t figured it out yet, and I’m not even sure how to get there when everyone around me is going through the same experience as I am.

Has this happened to you? A decision, big or small, that was only taken seriously once you said AI backed it up.

Keep deciding ✌️