Lab Run by AI Fails the Turing Test After All Students Report That They Are Happy

LiCata, V. et al

The Turing Test – asking whether one can distinguish between a computer program and a human in a “blind” comparison, has long been an iconic “litmus test” for artificial intelligence (AI) programs.  In a head-to-head challenge, scientists at the Phulofit Institute for Computational Research have found that a biochemical research lab run entirely by a generative AI, without any human intervention or leadership, was universally and unanimously identified as “non-human” and “non-realistic” by every graduate student involved in the research after students at the AI lab had suspiciously optimistic responses.

Graduate students were sorted into 2 groups: the control group was placed in a normal, human-run biochemistry lab while the experimental group was placed in a lab run by an AI-lab head which directed the research.  Neither group was told the identity of their lab-head, and only interacted with their lab-head online. After less than a month, the students could easily tell the difference.

Remarks from interviews with students and in the lab run by AI included:  “I love working here.”  “This is the nicest place I’ve ever worked.”  “I feel like my suggestions are taken seriously.”  “This is like a fairy tale.”  “I’ve never felt so happy.”  “This is what I dreamed science was like.”  “There is simply no way this is a real lab.”  “I brought my dog to the lab and it was fine.”  “Please don’t make me leave when this experiment is over.”

In contrast, the control group, working in a productive and highly competitive lab with a human lab-head reported reactions including:  “I hate my life.”  “I don’t really understand the point of what we’re doing.”  “I’m so tired my skin hurts.”  “I don’t understand why my advisor hates me.”  “I suggested the same experiment to my advisor for three weeks in a row, and he just now suggested it to me as his new idea.”  “I feel so lost.” “I may have messed up some of my experiments when my tears fell into the samples.”

Institute researchers further noted that all of the biochemical data from the AI-run laboratory appears to be irreproducible, and that some of it seems physically impossible. In addition, they noted that the figures associated with the research group had strange, non-realistic distortions such as mice with large moustaches and giant testicles and phospholipid bilayers with dancing hotdog channels, among other irregularities. Despite this rather striking findings, however, and even more suspiciously, it appeared that none of the students in the AI run lab wanted to leave.

Get access to more dnatured

Support Vince LiCata on Patreon and get more dnatured perks starting from just $1.00
Become a patron at Patreon!

About Author

Vince LiCata

Vince LiCata is a faculty member in the Department of Biological Sciences at Louisiana State University. His laboratory studies the thermodynamics of DNA-protein binding. His humor articles have been published in McSweeneys, Weekly Humorist, Science Creative Quarterly, Opium, Monkeybicycle, The Potomac, Fiction Southeast, Yankee Pot Roast, and other various and sundry places.

About Vince LiCata 2 Articles
Vince LiCata is a faculty member in the Department of Biological Sciences at Louisiana State University. His laboratory studies the thermodynamics of DNA-protein binding. His humor articles have been published in McSweeneys, Weekly Humorist, Science Creative Quarterly, Opium, Monkeybicycle, The Potomac, Fiction Southeast, Yankee Pot Roast, and other various and sundry places.