Brains respond differently to artificial intelligence and human speech

Humans aren’t very good at distinguishing human voices from those generated by artificial intelligence (AI), but our brains respond differently to human and AI voices, according to research presented today (Tuesday) at the Federation of European Neuroscience Societies (FENS). Forum 2024.

The study was presented by PhD student Christine Skjegstad and conducted by Ms Skjegstad and Professor Sascha Frühholz, both from the Department of Psychology at the University of Oslo (UiO) in Norway.

We already know that AI-generated voices have become so advanced that they are almost indistinguishable from the voices of real people. It is now possible to clone a person’s voice after just a few seconds of recording, and fraudsters have used this technology to imitate a loved one in distress and trick victims into sending money. While machine learning experts are developing technological solutions to detect AI voices, much less is known about how the human brain responds to these voices.”

Christine Skjegstad, Department of Psychology, University of Oslo

The study involved 43 people who were asked to listen to human and AI-generated voices expressing five different emotions: neutrality, anger, fear, joy, pleasure. [2]. They were asked to identify the voices as synthetic or natural while their brains were examined using functional magnetic resonance imaging (fMRI). fMRI is used to detect changes in blood flow in the brain, indicating which parts of the brain are energetic. Participants were also asked to rate the characteristics of the voices they heard in terms of naturalness, credibility and authenticity.

Participants correctly identified human voices only 56% of the time and artificial intelligence voices 50.5% of the time, meaning they were equally indigent at identifying both types of voices.

People were more likely to correctly identify a “neutral” AI voice as AI (75% compared to 23% who could correctly identify a neutral human voice as human), suggesting that people assume that neutral voices are more like AI. AI-neutral female voices were correctly identified more often than AI-neutral male voices. For elated human voices, the correct identification rate was 78% compared to just 32% for elated AI voices, suggesting that people associate happiness with something more human.

Both artificial intelligence and human neutral voices were perceived as the least natural, trustworthy and genuine, while human elated voices were perceived as the most natural, trustworthy and genuine.

However, by analyzing brain imaging, the researchers found that human voices evoked stronger responses in brain regions associated with memory (right hippocampus) and empathy (right inferior frontal gyrus). The AI ​​voices evoked stronger responses in areas related to error detection (right anterior medial cingulate cortex) and attention regulation (right dorsolateral prefrontal cortex).

Ms Skjegstad said: “My research shows that we are not very exact in recognizing whether a voice is human or AI-generated. Participants also often expressed how hard it was for them to distinguish the voices. This suggests that current artificial intelligence voice technology can mimic the human voice to such an extent that it is hard for humans to tell them apart.

“The results also point to a perception bias: neutral voices were more likely to be identified as AI-generated, and elated voices were more likely to be identified as more human, regardless of whether they actually were. This was especially true for neutral female AI voices, which may be because we’re familiar with female voice assistants like Siri and Alexa.

“Although we are not very good at distinguishing human voices from AI voices, there appears to be a difference in the brain response. “AI voices can induce heightened alertness, while human voices can evoke a sense of relatedness.”

The researchers now plan to investigate whether personality traits, such as extroversion or empathy, make people more or less sensitive to the differences between human voices and those of artificial intelligence.

Professor Richard Roche is the Chair of the FENS Forum Communications Committee and Deputy Head of the Department of Psychology at Maynooth University in Maynooth, Co. Kildare, Ireland, and was not involved in the research. He said: “Studying how the brain responds to AI voices is crucial as the technology continues to evolve. This research will lend a hand us understand the potential cognitive and social implications of AI voice technology, which can support policies and ethical guidelines.

“The threat of this technology being used to deceive and deceive people is obvious. However, there are also potential benefits, such as providing voice replacement for people who have lost their natural voice. Artificial intelligence voices can also be used in the therapy of some mentally ill people. health conditions.”