Close
today i learned (til)

The Uncanny Valley – An Exploration of How Humans Perceive AI

November 14, 2025

Have you ever seen a humanoid design that looks just off? Whether it’s the subtle placement of facial features, an abnormal glint in the eyes, or even just a subconscious lurking feeling that something’s not right, you’ve likely experienced the sensation often described as the uncanny valley.  

 

What is the uncanny valley?

The concept of the uncanny valley was first introduced in the early 1970s by a robotics professor, Masahiro Mori. Mori describes the term as an irregular representative of a linear relationship, stating, “as robots appear more humanlike, our sense of their familiarity increases until we come to a valley.” (Mori et al., 1970) This ‘valley’ he refers to is a point in the relationship between human ‘likeness’ and familiarity in which familiarity plummets before increasing again. Mori designed a graphical relationship indicating this phenomenon, which can be viewed in the image below. This dynamic can be even further exacerbated by attentional human-like cues added, such as movement or touch sensation (Mori et al., 1970). 

 

 

(Original diagram created by Masahiro Mori, who coined the uncanny valley term in the early 70s.) Source: Wikimedia Commons [https://commons.wikimedia.org/w/index.php?curid=2041097]

 

The actual origin of the uncanny valley is a question sought after and theorized on by many researchers. A few common hypotheses about its origin have been proposed though. Some examples that will be further discussed below are categorization uncertainty and perceptual mismatch. Unfortunately these hypotheses and actual root of the phenomenon is under researched area, making it difficult to tie a definitive cause (Kätsyri et al., 2025). 

 

Categorical perception deals with one’s predisposition to better perceive something if it distinctly falls into a mental category. This causes things felt to fall into discrete categories to be highlighted in their uniqueness, and those that remain similar to have an understated emphasis on difference (Goldstone & Hendrickson, 2009). The hypothesis is that a cognitive dissonance is created when the uncanny valley is reached in a humanoid design. Essentially, at this stage, there is a difficulty in the mental categorization process, with ambiguity in the design making it hard to determine if the figure falls into the ‘human’ category or the ‘robot/inhuman’ category. This uncertainty is thought to bring the discomforting feeling described by the effect (Kätsyri et al., 2025).

 

Another hypothesis is rooted in the idea of perceptual mismatch. This theory follows a similar line of reasoning to the categorization uncertainty hypothesis, being that the uncanny valley effect is thought to be brought on by a feeling of discomfort that arrives by the brain’s difficulty to discern something. In this case, it relates to the perceptual confusion brought along by something with high human likeness with a feature that is inhuman or artificial in appearance. The discrepancy between a primarily human perception and an identifying feature that is misaligned with that is thought to create the uncomfortable feeling of uncanniness (Kätsyri et al., 2025). 

 

Uncanny Valley and AI

As time has gone on, the general public has witnessed the rapid development of technology. With our world’s rising implementation of artificial intelligence (AI) run programs, generative technologies, and interfaces, the question arises, how does this concept intersect with AI? Can the uncanny valley phenomenon expand beyond visual design?

 

A recent study explores these perceptions in action and how it affects user interaction and emotional responses. Deepali Kishnani, a student at MIT, employed an experimental structure that examined text interactions between participants and 3 variables – a bot engineered to be ‘human-like’, a bot engineered to be ‘uncanny-like’, and a human for control. After conversing with one of the 3, the participant ranked the system on anthropomorphism, animacy, likeability, perceived intelligence, and safety (Kishnani, 2025). 

 

Based on the rating and analysis of the results, it was concluded that there is evidence that the uncanny valley is present in text-based interaction, as the ‘uncanny-like’ bot was consistently rated lowest on all dimensions. Findings also revealed that people are having an increasingly more difficult time distinguishing human interaction from AI interactions, as 60% of participants in the human control group incorrectly identified their chat companion as AI (Kishnani, 2025).

 

With the now confirmed presence of possible uncanny valley in AI chatbot interactions, an additional question of how it influences trust from users arises. Another study explores this concept, seeking how the anthropomorphism of AI systems induces concern into the patron, instigating fear of human-natured negative actions onto the inhuman bot (Kim et al., 2024). This study specifically introduces a concept known as ‘uncanny valley of the mind,’ which focuses more on human-like intentions rather than human-like appearance. They suggest that the uncanny valley of the mind is brought on when a chatbot is perceived to have human-like mentalites (also known as a ‘digital mind’), they will also possess human-like intentions- such as the ability to misuse information (Kim et al., 2024).

 

To test this phenomenon, researchers created 2 chatbots that could be accessed through facebook messenger, one ‘anthropomorphized’ and one ‘non-anthropomorphized.’ Through a series of interactions, the bot asked for personal information and later synthesized either an ad containing that information or a generic ad without it. Interestingly, when the anthropomorphized bot presented the user with a personalized ad, participants reported increased privacy concerns. This effect was not present in the non-anthropomorphized ads, regardless of it being the personalized or non-personalized ad. With this being said, it was determined that higher anthropomorphism in chatbots decreases consumer trust/likelihood to purchase a product when personal information is included in the pitch (Kim et al., 2024).

 

What Does this Mean for the Future of AI?

It seems that as the days go on, more applications of AI are created and present in a multitude of functions. As implementations into features like writing programs, customer service, and ad-personalization grows, opportunity for distrust in these systems increases along with it. The research above proves that the uncanny valley can seep into perceptions of these AI powered systems, so how will producers handle this realization?

 

In my opinion, there are two ways this industry can progress to mediate this issue from a psychological standpoint. Either producers will make efforts to develop AI systems that are less uncanny and more perceptual humans or consumers will begin to gravitate away from trust. Historically both of these things have occurred in product development and examples regarding the uncanny valley can even be observed. 

 

A great exemplification of both of these mediations is the design of ‘Sonic the Hedgehog’ in the 2020 film rendition. When the first trailer released, fans of the franchise were horrified by the uncanny design of their beloved main character. Sonic, a very cartoonish character, had been adapted to his life action role by combining realistic CGI texturing and a flurry of human-like features on a very non-human creature. The uproar after the release was instant, with many fans taking to social media to detest the movie and state their refusal to endorse it. (Alexander, 2020) This is an example of the consumers gravitating away from trust. Producers were then forced to change the design to be more appealing and cartoonish, exemplifying the need for producer design in a way that is more appealing to consumers. 

 

The true outcome of how AI develops can only be told by time. Until then, be a critical consumer. Do you find yourself experiencing a lurking, unsettling, off feeling when chatting with AI systems?

 

References:

Alexander, B. (2020, February 17). How “Sonic the Hedgehog” avoided death by Twitter with a dramatic, game-saving redesign. USA Today. https://www.usatoday.com/story/entertainment/movies/2020/02/13/sonic-the-hedgehog-movie-redesign-how-backlash-made-character-better/4730593002/

Goldstone, R. L., & Hendrickson, A. T. (2009, December). Categorical perception. WIREs Cognitive Science. https://wires.onlinelibrary.wiley.com/doi/10.1002/wcs.26

Kim, W., Ryoo, Y., & Choi, Y. K. (2024). That uncanny valley of mind: When anthropomorphic ai agents disrupt personalized advertising. International Journal of Advertising, 1–30. https://doi.org/10.1080/02650487.2024.2411669

Kishnani, D. (2025, March 7). The uncanny valley: An empirical study on human perceptions of AI-generated text and images. The Uncanny Valley: An Empirical Study on Human Perceptions of AI-Generated Text and Images. https://dspace.mit.edu/handle/1721.1/159096

Kätsyri, J., Förger, K., Mäkäräinen, M., & Takala, T. (2025, November 1). A review of empirical evidence on different uncanny valley hypotheses: Support for perceptual mismatch as one road to the valley of eeriness. Frontiers. https://www.frontiersin.org/journals/psychology/articles/10.3389/fpsyg.2015.00390/full#B19

Mori, M. (1970). The Uncanny Valley. https://www.almendron.com/tribuna/wp-content/uploads/2018/01/morunc.pdf 

0Likes
About Author

Chloe is an undergraduate 2nd year Psychology major and Industrial Design minor. She has a particular interesting in human-factors design, engineering psychology, and behavioral analysis. In her free time, Chloe loves listening to punk and alternative rock, reading and watching any form of media, and baking.

You May Also Like

today i learned (til)
Going Crazy in Quarantine? You’re Not Alone!
today i learned (til)
Noise Pollution Effect on Sleep
four blind boxes on a dark purple table with sparkles in the background. The theme is magical space.
today i learned (til)
Blind Box Culture and Its Appeal
today i learned (til)
Can Algorithms Deliver Justice?

Your journey starts here! Let’s talk!

ThemeRex © 2026. All rights reserved.
  Timothy Durham was convicted of rape when a test showed his DNA genotype matched the DNA recovered from the crime scene. Eye witnesses testified that Durham was in a different state at the time of the incident, but he wasn’t exonerated from his 3000 year sentence until follow up DNA analysis showed that the preliminary forensic analysis used in the trial was misinterpreted (Thomas et al.). DNA testing serves an invaluable part in our criminal justice system, but it comes with complications and bias that can lead to wrongful convictions and misinformed juries.  

Background:

Forensic genetics aims to identify the origin of a biological sample. It can be traced back to Karl Landsteiner who discovered human ABO blood groups. With blood groups, suspects could be eliminated based on their blood type. For example, if a suspect had type A blood and there was type B blood left at the scene by the perpetrator, the investigators could rule out the suspect. Unfortunately, the reverse didn’t work. If the suspect and perpetrator both had the same blood type, it can’t be concluded that the suspect was guilty. Blood identifiers evolved into 8 distinct categories which improved the ability to eliminate suspects, but could only conclude that (if the system matched) it might be the perpetrator or this person might be related to the perpetrator (Jobling and Gill). The field evolved in the 1980s with the discovery of Minisatellites, lines of repetitive DNA that don’t encode proteins. Minisatellite patterns in DNA create what we know today as the DNA fingerprint, and it has a much higher accuracy rate than previous methods (Augustyn et al.). Around the same time, a breakthrough for victims of sexual assault was discovered. In the case of assault, bodily fluids of both the victim and the assaulter can get mixed. A process called differential lysis was created which could selectively increase the concentration of sperm in a fluid sample thereby allowing for accurate DNA testing (Jobling and Gill). Forensics evolved to use single-locus probe testing after several years because of its ease of interpretation. It looked at specific minisatellites that had high variability between subjects. This was the first DNA-based method to be used in a criminal investigation. Two murder homicides from several years apart were connected using DNA evidence. The evidence was then used to exonerate an innocent man, and the guilty party was found when he tried to evade mandatory DNA testing for all men in the small town. His guilt was then reinforced when the DNA profiles matched, and he gave a confession (Jobling and Gill). Finding DNA at a crime scene doesn’t guarantee that you’ll find a murderer. There are issues with non-pure samples and limited samples that make DNA work difficult, and accreditation of facilities and new techniques and take time (Jobling and Gill).  

Subjectivity of DNA Interpretation:

In a 2009 publication, the National Academy of Science discusses the possibility for subjective influence in lab based forensics. The interpretation bias happens in complex cases where, for example, there could be a mixture of DNA. A study published in 2011 looked specifically at this issue. A gang rape happened in Georgia, and the mixture in DNA from the case was used to prosecute one suspect after another suspect took a plea bargain to testify against him. With DNA mixtures, there are three possible conclusions: suspect excluded, suspect cannot be excluded, and inconclusive. If the suspect is excluded, it means that there’s little to no possibility that his DNA is part of the sample. If the suspect cannot be excluded, it means that there’s a possibility that his DNA is part of the sample. In the trial, the DNA examiners who knew the context of the case concluded that the suspect being prosecuted could not be excluded. The 2011 study gave the DNA evidence from this case and the DNA profiles of the suspects to 17 forensic examiners. These examiners had no context of the case, and they all worked independently. Of the 17, only 1 came to the same conclusion (suspect cannot be excluded) as the examiners who knew the context of the case. Twelve of the 17 concluded that the suspect who was prosecuted and convicted had no relation to the DNA found in the sample. Not only does this show that subjectivity occurs due to the context of the case, but the 17 examiners came up with different findings based on their past experiences meaning that not all DNA analysis is objective. Some organizations like the Scientific Working Group on DNA Analysis Methods say that quantitative statistical evidence should be presented with every categorical conclusion, but the US doesn’t require this (Dror and Hampikian).

False Positives:

There are two categories that cause wrongful identification in a DNA report: coincidental matches and false positives. Coincidental matches are DNA matches between two unrelated people. The probability of these matches is about 1 to 500,000 for white individuals and about 1 to 350,000 for black individuals. This small probability of coincidental matches makes people feel confident in DNA evidence. False positives, on the other hand, are matches between DNA and test DNA that result from issues in the forensics lab. There have been multiple cases of forensics labs accidentally switching victim and suspected rapist DNA which lead to convictions that were later overturned when the lab realized the error. There are also cases of misreading or misunderstanding a sample. In a case from 1993, a man was convicted of rape because someone interpreted a mixed sample of victim and perpetrator DNA as an individual sample. The conviction was again overturned. To prevent false convictions, retesting DNA is allowed, but it’s not an efficient solution. Many people are pushing for statistical information on the probability of false positives, so that the jurors can make accurate judgements based on DNA evidence (Thomas et al.).  

Conclusion:

DNA analysis is a useful tool for the criminal justice system, but subjectivity can be an issue in forensic testing. To combat wrongful convictions, there are organizations like The Innocence Project who use DNA testing to free the innocent, but reformed forensic testing standards should be put in place to prevent subjectivity and to ensure juries are correctly informed during trials. https://www.innocenceproject.org/    

Citation:

Augustyn , Adam, et al. “DNA Finerprinting.” Encyclopedia Britannica , 26 Apr. 2018, www.britannica.com/science/DNA-fingerprinting#ref153186. Dror, Itiel E., and Greg Hampikian. “Subjectivity and Bias in Forensic DNA Mixture Interpretation.” Science & Justice, vol. 51, no. 4, 2011, pp. 204–208., doi:10.1016/j.scijus.2011.08.004. Jobling, Mark A., and Peter Gill. “Encoded Evidence: DNA in Forensic Analysis.” Nature Reviews: Genetics, vol. 5, no. 10, Oct. 2004, pp. 739–751., doi:10.1038/nrg1455. Thompson, William C, et al. “How the Probability of a False Positive Affects the Value of DNA Evidence.” Journal of Forensic Sciences, vol. 48, no. 1, Jan. 2003, pp. 1–8., doi:10.1520/jfs2001171.