From left, Hong Wan, KAIST doctoral pupil, and Professor Choi Moon-jung. Provided by KAIST
■ KAIST introduced on the twenty eighth {that a} analysis workforce led by Professor Choi Moon-jung on the Graduate School of Science and Technology Policy has quantitatively analyzed that the responses of the generative synthetic intelligence mannequin “ChatGPT-4o” embed subtle age stereotypes towards older adults. The findings had been printed in the February 2026 particular problem of the worldwide gerontology journal The Gerontologist. The workforce collected 900 texts by utilizing impartial prompts that requested the mannequin to explain the traits of age teams in 10-year intervals from 10 to 90 years previous, and utilized the Stereotype Content Model (SCM) to investigate the scale of “warmth” and “competence.” The evaluation confirmed that for teams aged 60 and older, scores for heat—comparable to kindness and consideration—had been excessive, whereas competence—comparable to capacity and experience—tended to be expressed at decrease ranges than for youthful age teams. In addition, as age elevated, expressions associated to assertiveness, which mirror confidence and management, decreased, indicating that AI can painting older adults as clever and benevolent whereas depicting them as having decrease company and proactiveness. The analysis workforce identified that repeated publicity to such expressions could reinforce social prejudice towards older adults and result in digital ageism.
■ Ulsan National Institute of Science and Technology (UNIST) introduced on the twenty sixth {that a} analysis workforce led by Professor Yoon Seonghwan on the Graduate School of Artificial Intelligence has recognized, from the attitude of loss panorama flattening, why multimodal synthetic intelligence (AI) learns extra precisely and stably than single-modal AI. The outcomes can be offered on the International Conference on Machine Learning (ICML 2026), a number one worldwide convention in AI. Multimodal studying is a technique in which AI makes use of completely different types of information collectively—comparable to photographs, speech, and textual content—to higher perceive the identical object or state of affairs. The workforce defined that when information from a number of modalities are discovered collectively, the loss panorama turns into flatter, thereby enhancing robustness in order that the mannequin can reply stably even to conditions it has not encountered throughout coaching. They additionally mathematically launched a “convolutional smoothing effect,” in which completely different information varieties mitigate tough variations in error as if averaging them out, and proposed distribution-based multimodal studying (DML), which re-pairs completely different modality information inside the similar goal class. Researcher Lee Jaejun of the UNIST Graduate School of Artificial Intelligence participated as first creator, and the workforce reported that DML outperformed standard fixed-pair studying in picture–textual content retrieval and classification experiments.
■ Gwangju Institute of Science and Technology (GIST) introduced on the twenty eighth that it held the “GIST Integration and Innovation Forum” on the GIST Oryong Hall on June 26 and offered a imaginative and prescient for a science and technology-based innovation ecosystem in preparation for the period of the built-in Jeonnam–Gwangju Special City. The discussion board introduced collectively about 100 individuals from GIST and business, academia, analysis institutes, and authorities to debate regional progress methods for the age of AI transformation (AX) and wide-area integration. On the identical day, GIST additionally held the opening ceremony for the Startup Innovation Promotion Center and signed a memorandum of understanding with the Chonnam National University Industry-Academic Cooperation Foundation and the KENTECH Value Creation Center of the Korea Institute of Energy Technology (KENTECH) to advertise regional innovation, know-how startups, and commercialization. The Startup Innovation Promotion Center will function as a devoted group that gives built-in assist for commercialization of research成果, know-how startups, funding linkage, and business–academia cooperation. At the discussion board, Professor Kim Jae-kwan of the Department of Biomedical Science and Engineering, Dean Kim Kanguk of the School of Information and Communications, Professor Park Kihong of the Department of Environmental and Energy Engineering, and Special Professor Kwak Jaewon of the AI Graduate School of Policy and Strategy gave shows on subjects comparable to AI semiconductors, vitality, deep-tech startups, and visions for the coexistence of people, AI, and the Earth. GIST President Lim Ki-cheol acknowledged, “By using the Startup Innovation Promotion Center as a hub to connect research, startups, and industry, we will take the lead in enabling Gwangju and Jeonnam to emerge as a core driver of Korea’s future growth.”
– doi.org/10.1093/geront/gnaf291
Copyright ⓒ DongA Science. All rights reserved.