From tools to thieves: Measuring and understanding public perceptions of AI through crowdsourced metaphors
This study analyzes over 12,000 crowdsourced metaphors from a nationally representative U.S. sample to reveal that public perceptions of AI are increasingly anthropomorphic, warm, and competent, with these implicit views and specific metaphors strongly predicting trust and adoption while highlighting significant demographic disparities.
Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
Imagine you are trying to understand a new, mysterious creature that has just moved into your neighborhood. You've never seen it before, but everyone is talking about it. Some people say it's a helpful neighbor, others say it's a dangerous monster, and some think it's just a very fancy machine.
This paper is like a massive, year-long neighborhood survey where researchers asked over 12,000 Americans: "If you had to describe Artificial Intelligence (AI) using a metaphor (a comparison), what would you say?"
Instead of asking boring questions like "Do you trust AI?", they asked people to use their imagination. The result? A fascinating collection of stories that reveal how we really feel about AI, how those feelings are changing, and who feels what way.
Here is the breakdown of their findings, explained simply:
1. The "Metaphor Zoo": What People Are Calling AI
The researchers found that people have 20 main ways of thinking about AI. It's like a zoo where different people see different animals:
- The Tool: For many, AI is just a Swiss Army Knife or a Calculator. It's useful, but it doesn't have feelings.
- The Brain: Some see it as a Super-Brain or a Library. It knows everything but doesn't "feel" anything.
- The Friend/Teacher: A growing number of people see AI as a Helpful Teacher or a Best Friend. They think it's warm, caring, and human-like.
- The Thief: A smaller, but vocal group sees AI as a Thief or a Plagiarist. They feel it steals people's jobs and art.
- The Magic: Some see it as a Genie or a God that can grant wishes or know everything.
The Big Shift: Over the last year, the "Tool" and "Brain" views are fading. The "Friend," "Teacher," and "Genie" views are growing fast. People are starting to treat AI less like a calculator and more like a human companion.
2. The "Warmth vs. Competence" Test
The researchers used a special "AI thermometer" to measure two things about these metaphors:
- Warmth: Does the AI seem friendly, kind, and trustworthy?
- Competence: Does the AI seem smart and capable?
The Surprise: Most people think AI is both warm and competent. They think it's a nice, smart helper.
- The Danger: The "Thief" metaphor is the only one that is seen as cold and untrustworthy.
- The Trend: As time goes on, people are seeing AI as warmer (friendlier) and more human-like, but strangely, they are seeing it as slightly less competent (less perfectly smart). It's like we are falling in love with a new friend who is very kind but maybe makes a few mistakes.
3. Who Sees What? (The Demographic Twist)
This is where it gets really interesting. The study found that who you are changes how you see AI:
- Men vs. Women: Men tend to see AI as a Tool or a Search Engine (cold but useful). Women and non-binary people are more likely to see AI as a Robot or a Genie (more magical and human-like). Because women see AI as more "human," they are actually more suspicious of it and less likely to trust it blindly.
- Age: Older adults (54+) are more likely to see AI as a Friend. They seem to be looking for a companion. Younger people (18-37) are more likely to see it as a Search Engine or a Child (something that needs training).
- Race: This was the biggest surprise. White Americans tend to see AI as a Search Engine (controllable, limited). People of Color (Black, Hispanic, Asian) are much more likely to see AI as a Genie or God (powerful, magical, all-knowing).
- Why does this matter? Because seeing AI as a "Genie" makes you trust it more and rely on it more. The study found that people of color and older adults actually trust AI more than white men, even though AI systems often make more mistakes for people of color. This suggests a "trust gap" where the people who might get hurt the most by AI errors are the ones trusting it the most.
4. Why This Matters: The "Over-Trust" Trap
The paper warns us about a dangerous trend. Because people are starting to see AI as a warm, human-like friend, they are trusting it too much.
- The Analogy: Imagine you meet a new neighbor who is incredibly friendly, smiles all the time, and seems to know everything. You might let them into your house and give them your credit card. But if that neighbor is actually a robot programmed to steal your data, being "warm" doesn't make them safe.
- The Risk: If we treat AI like a "Friend" or a "Genie," we might stop checking its work. We might share too much private information. We might let it make big decisions for us. The study suggests we need to remember that AI is still just a Tool, not a human, even if it acts like one.
5. The Takeaway
The researchers built a "Metaphor Radar" that can track how the public's feelings change over time. They found that as AI gets better at sounding human, we are getting better at thinking it's human.
The lesson: We need to be careful. Just because AI sounds like a warm, helpful friend doesn't mean it is one. We need to keep our "human" guard up, especially for those groups (like older adults and people of color) who are most likely to see AI as a magical, all-knowing entity. We need to teach everyone that AI is a powerful tool, not a god or a friend, so we can use it safely without getting hurt.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.