Scientists and an IT company were unable to create an artificial intelligence (AI) model that would be able to correctly distinguish between human emotions. This was reported by The Conversation, Day.Az reports with reference to Lenta.ru.
Media journalists have noticed that AI services are appearing on the market, the creators of which promise that they can distinguish and work with human emotions. However, scientists believe that neural networks and machines are still inferior to humans in this skill.
As an example, the authors cited the Australian startup inTruth, whose specialists stated that their AI model is capable of distinguishing human emotions and drawing conclusions based on this. However, journalists noted that there is still no scientific evidence that machines can understand a person’s emotional state.
AI has previously been found to discriminate on the basis of race, gender and disability. In one case, the model perceived black people’s emotions as more aggressive. Experts also note that the studied AI models tend to rely on physiognomy and phrenology, pseudoscientific areas recognized by the community, when determining emotions.
inTruth countered that they had achieved a breakthrough in identifying human emotions. Experts noted that the problem of discrimination or bias was not the fault of the technology itself, but because of the set of data on which the AI was trained.