Key points are not available for this paper at this time.
Background Recent studies suggest the use of images generated by artificial intelligence (AI) should be promoted among researchers and clinicians.Aims The current study aimed to examine the naming performances of neurotypical individuals through images generated by Bing Image Creator (BIC) that uses DALL-E 3 system according to a number of variables including demographic, word classes, and psycholinguistic parameters. To fulfil these aims, two separate but interconnected studies were conducted.Methods & Procedures Prior to Study 1, the images of 81 nouns and 52 actions were formed and reviewed as a result of entering the relevant prompts for these stimuli in BIC. The team agreed on the depiction of 71 nouns and 48 actions. Afterwards, 50 third- or fourth-year undergraduate students rated these nouns and verbs according to the parameters of naming agreement, image agreement, and visual complexity. The students responded to two open-ended questions examining the views about AI-generated images, which was analyzed through thematic analysis. The administration was conducted via QualtricsTM. In Study 2, the final list was administered to 152 neurotypical individuals via QualtricsTM.Outcomes & Results In Study 1, the findings showed that four nouns and two verbs should be removed from the list, which was updated, including 67 nouns and 46 actions for final administration in Study 2, and four codes were generated following thematic analysis. In Study 2, the findings showed the following: Regarding demographic groups, the participants in the younger group scored higher in noun and verb naming than the elder group. As regards latency, younger adults named nouns and verbs faster than the older ones. The noun naming scores of those in the graduate group scored significantly higher than those in the remaining two groups. The verb naming scores of those in the graduate group scored significantly higher than those with associate degrees and below. Regarding latency, those receiving the lowest education spent more time in naming nouns and verbs compared to the remaining education groups. Regarding word class parameter, participants were more successful in naming nouns compared to verbs. The nouns were named faster than the verbs. Significant correlations were obtained between naming accuracy and latency, and the psycholinguistic variables. The findings showed the AI-generated images were successful in assessing the naming skills of participants. The use of AI in image generation could be promising in the field of Speech and Language Therapy, especially in the preparation of individualized therapy materials.
Özdemir et al. (Fri,) studied this question.