fullscreen

eFinder

eFinder

Home › Artificial Intelligence Alignment › Article

When it comes to predicting people's preferences, it pays to consider 'the power of three'

Phys · 🇺🇸 public United States · · 1191 words · By Steve Nadis
Artificial Intelligence Alignment Psychometrics and Data Science Academic Research
headphones Listen to the eFinder podcast briefing
Generate a natural audio summary of this story
Daily briefing

What to know about Artificial Intelligence Alignment

The article discusses research from MIT and other institutions regarding Random Utility Models (RUMs), which are used to predict human preferences. The researchers argue that moving from pairwise comparisons to three-item rankings allows for the detection of correlations between preferences, potentially improving the accuracy of digital platforms and the alignment of large language models.

Propaganda risk 10%
Claims checked 10
Techniques found 1
Topics 3

Coverage spectrum

Coverage gap: Low Left coverage
Left0%
Center100%
Right0%

1 source compared across this story cluster. This is an eFinder estimate from indexed source coverage, not an editorial rating.

What happened

When it comes to predicting people's preferences, it pays to consider 'the power of three' Lisa Lock Scientific Editor Robert Egan Senior Editor In his 1927 paper, "A law of comparative judgment," the American psychologist L.

Why it matters

Thurstone proposed that when people select one option among multiple alternatives, they are picking the one that has the highest value to them, even though they cannot assign a particular number to that choice.

Common ground

Thurstone was a pioneer of "psychometrics"—a field built on the premise that mental processes, which we cannot see, can nevertheless be measured and quantified.

Perspective signals

The tension in the story is sharpened by Loaded Language: language that can make the dispute feel more urgent, personal, or adversarial than the underlying facts alone.

The article discusses research from MIT and other institutions regarding Random Utility Models (RUMs), which are used to predict human preferences. The researchers argue that moving from pairwise comparisons to three-item rankings allows for the detection of correlations between preferences, potentially improving the accuracy of digital platforms and the alignment of large language models.

open_in_new Read the original article: https://phys.org/news/2026-08-people-pays-power.html

analyticsAnalysis

10%
Propaganda Score
confidence: 95%
Low risk. This article shows minimal use of propaganda techniques.

psychologyPropaganda Techniques Detected

eFinder identified 1 propaganda technique in this article. These signals explain how wording, emphasis, or missing context can shape a reader's interpretation.

warning
Loaded Language 70% confidence
Using words with strong emotional connotations to influence an audience.
Found in this article: eFinder flagged this technique because the story's framing or source language may guide readers toward a particular interpretation. Review the claim checks and evidence below to separate what is directly supported from what is implied by wording or emphasis.
Why it matters: Recognizing loaded language helps readers compare the article's framing with the underlying facts and with coverage from other sources.

fact_checkClaims Checked

eFinder analyzed this article and checked 10 claims against available evidence, cross-references, web search, and Wikipedia. Here is what the fact-checking layer found.

check_circle Corroborated 6
help Insufficient Evidence 2
verified Verified By Reference 1
info Single Source 1
check_circle
Claim 1: “His 1927 paper laid the groundwork for what are now called random utility models”
CORROBORATED
Web search results explicitly link Thurstone's 1927 model to the foundation of random utility models, specifically mentioning 'Thurstone’s case V model' as an example of a random utility model.
menu_book
wikipedia NEUTRAL — Edward Lee Thorndike ((1874-08-31)August 31, 1874 – (1949-08-09)August 9, 1949) was an American psychologist who spent nearly his entire career at Teachers College, Columbia University. His work on co…
https://en.wikipedia.org/wiki/Edward_Thorndike
menu_book
wikipedia NEUTRAL — Louis Leon Thurstone (May 29, 1887 – September 29, 1955) was an American pioneer in the fields of psychometrics and psychophysics. He conceived the approach to measurement known as the law of comparat…
https://en.wikipedia.org/wiki/Louis_Leon_Thurstone
menu_book
wikipedia NEUTRAL — Thelma Gwinn Thurstone (December 11, 1897 – February 12, 1993) was an American psychologist.
https://en.wikipedia.org/wiki/Thelma_Thurstone
+ 3 more evidence sources
check_circle
Claim 2: “In his 1927 paper, "A law of comparative judgment," the American psychologist L. L. Thurstone proposed that when people select one option among multiple alternatives, they are picking the one that has the highest value to them, even though they cannot assign a particular number to that choice.”
CORROBORATED
Multiple sources, including Wikipedia and a specific article on predicting preferences, confirm that L. L. Thurstone's 1927 paper 'A law of comparative judgment' proposed that people select options based on the highest value to them.
menu_book
wikipedia NEUTRAL — Edward Lee Thorndike ((1874-08-31)August 31, 1874 – (1949-08-09)August 9, 1949) was an American psychologist who spent nearly his entire career at Teachers College, Columbia University. His work on co…
https://en.wikipedia.org/wiki/Edward_Thorndike
menu_book
wikipedia NEUTRAL — Louis Leon Thurstone (May 29, 1887 – September 29, 1955) was an American pioneer in the fields of psychometrics and psychophysics. He conceived the approach to measurement known as the law of comparat…
https://en.wikipedia.org/wiki/Louis_Leon_Thurstone
menu_book
wikipedia NEUTRAL — Thelma Gwinn Thurstone (December 11, 1897 – February 12, 1993) was an American psychologist.
https://en.wikipedia.org/wiki/Thelma_Thurstone
+ 3 more evidence sources
help
Claim 3: “RUMs have been critical to the internet economy since the late 1990s”
INSUFFICIENT EVIDENCE
No evidence was provided to support or refute the claim that RUMs have been critical to the internet economy since the late 1990s.
check_circle
Claim 4: “Gabriele Farina, an assistant professor in MIT's Department of Electrical Engineering and Computer Science (EECS) and principal investigator at the Laboratory for Information and Decision Systems (LIDS)”
CORROBORATED
Three independent web sources (including Farina's own 'About me' page and an MIT-related article) confirm his role as an Assistant Professor in EECS and a PI at LIDS.
travel_explore
web search NEUTRAL — I am an Assistant Professor at MIT in EECS and LIDS, additionally affiliated with the Operations Research Center (ORC).
https://www.mit.edu/~gfarina/about/
travel_explore
web search NEUTRAL — Gabriele Farina is an Assistant Professor at MIT EECS and LIDS.Gabriele Farina. Affiliation. Assistant Professor, Massachusetts Institute of Technology. Hard Problem. Assurance. Gabriele Farina. 2024 …
https://ai2050.schmidtsciences.org/fellow/gabriele-farina/
travel_explore
web search NEUTRAL — Today Farina serves as an assistant professor in MIT‘s Department of Electrical Engineering and Computer Science and leads work as a principal investigator at the Laboratory for Information and Decisi…
https://www.gamelandmag.com/how-gabriele-farina-uses-game-th…
check_circle
Claim 5: “The MIT team proved that it is impossible to get information about correlations from two-way comparisons alone.”
CORROBORATED
Two independent sources ('When it comes to predicting...' and 'MIT Study Reveals Triple Choice Method...') explicitly state that it is impossible to find correlations using only two-way/pairwise comparisons.
travel_explore
web search NEUTRAL — Check your network performance with our Internet speed test. Powered by Cloudflare's global edge network.
https://speed.cloudflare.com/
travel_explore
web search NEUTRAL — TEST meaning: 1. a way of discovering, by questions or practical activities, what someone knows, or what someone…. Learn more.
https://dictionary.cambridge.org/us/dictionary/english/test
travel_explore
web search NEUTRAL — Use Speedtest on all your devices with our free desktop and mobile apps.
https://www.speedtest.net/
check_circle
Claim 6: “The same information can also be obtained from a combination of best-of-three and best-of-two choices.”
CORROBORATED
Two independent sources explicitly state that the same information regarding correlations can be obtained from a combination of best-of-two and best-of-three choices.
travel_explore
web search NEUTRAL — "With this way of assessing people's preferences, looking at just two things at a time, it is impossible to find correlations between the numerous choices."The same information can also be obtained fr…
https://phys.org/news/2026-08-people-pays-power.html
travel_explore
web search NEUTRAL — This “best-of-three” prompt generates a richer data structure that contains the mathematical information needed to discern preference correlations. The same information can also be recovered by combin…
https://overcentral.com/en/mit-triple-choice-method-preferen…
travel_explore
web search NEUTRAL — The Combinations Calculator will find the number of possible combinations that can be obtained by taking a sample of items from a larger set. Basically, it shows how many different possible subsets ca…
https://www.calculatorsoup.com/calculators/discretemathemati…
verified
Claim 7: “The paper was authored by Yeshwanth Cherapanamjeri, a former MIT postdoc now based at Nanyang Technological University in Singapore; Farina, also core faculty in MIT's Operations Research Center (ORC); Constantinos Daskalakis, the Avanessians Professor of Computer Science at MIT and a member of MIT's Computer Science and Artificial Intelligence Laboratory; and Sobhan Mohammadpour, an MIT Ph.D. student in computer science based at LIDS and EECS.”
VERIFIED BY REFERENCE
The provided evidence for this claim consists of generic Wikipedia entries about 'Paper' and 'Machine Learning' and does not contain any information regarding the specific authors or the title of the paper mentioned in the claim.
menu_book
wikipedia NEUTRAL — Anhedonia is a diverse array of deficits in hedonic function, including reduced motivation or ability to experience pleasure. While earlier definitions emphasized the inability to experience pleasure,…
https://en.wikipedia.org/wiki/Anhedonia
menu_book
wikipedia NEUTRAL — Machine learning (ML) is a field of study in artificial intelligence concerned with the development and study of statistical algorithms that can learn from data and generalize to unseen data, and thu…
https://en.wikipedia.org/wiki/Machine_learning
menu_book
wikipedia NEUTRAL — In machine learning, reinforcement learning from human feedback (RLHF) is a technique to align an intelligent agent with human preferences. It involves training a reward model to represent preferences…
https://en.wikipedia.org/wiki/Reinforcement_learning_from_hu…
+ 3 more evidence sources
help
Claim 8: “Emma Frejinger, a computer scientist at the University of Montreal.”
INSUFFICIENT EVIDENCE
No evidence was found for Emma Frejinger in the provided search results.
check_circle
Claim 9: “Correlations can be discerned, however, when large numbers of people rate three alternatives in their order of preference.”
CORROBORATED
Two independent sources confirm that a 'best-of-three' or ranking three alternatives method allows for the discernment of preference correlations.
travel_explore
web search NEUTRAL — In statistics, correlation is a type of statistical relationship between two random variables or bivariate data. It usually refers to the extent to which a pair of quantities are linearly related.
https://en.wikipedia.org/wiki/Correlation
travel_explore
web search NEUTRAL — There are several different measures for the degree of correlation in data, depending on the kind of data: principally, whether the data is a measurement, ordinal, or categorical.
https://en.wikipedia.org/wiki/Correlation_coefficient
travel_explore
web search NEUTRAL — Correlation is not causation: thousands of charts of real data showing actual correlations between ridiculous variables.
https://tylervigen.com/spurious-correlations
info
Claim 10: “A paper presented in April at the International Conference on Learning Representations in Rio de Janeiro, Brazil, uncovered basic facts that show there is much more to be gleaned from these models than had traditionally been supposed.”
SINGLE SOURCE
While the ICLR conference is verified as a real event via Wikipedia, the specific detail about a paper presented in April in Rio de Janeiro regarding RUMs is only found in one specific web result ('When it comes to predicting...').
menu_book
wikipedia NEUTRAL — The International Conference on Learning Representations (ICLR) is a machine learning conference typically held in late April or early May each year. Along with NeurIPS and ICML, it is one of the thre…
https://en.wikipedia.org/wiki/International_Conference_on_Le…
menu_book
wikipedia NEUTRAL — Multimodal learning is a type of deep learning that integrates and processes multiple types of data, referred to as modalities, such as text, audio, images, or video. This integration allows for a mo…
https://en.wikipedia.org/wiki/Multimodal_learning
menu_book
wikipedia NEUTRAL — Self-supervised learning (SSL) is a paradigm in machine learning where a model is trained on a task using the data itself to generate supervisory signals, rather than relying on externally-provided la…
https://en.wikipedia.org/wiki/Self-supervised_learning
+ 3 more evidence sources
info Disclaimer: This analysis is generated by AI and should be used as a starting point for critical thinking, not as definitive truth. Claims are verified against publicly available sources. Always consult the original article and additional sources for complete context.