From 'Help' to Helpful: A Hierarchical Assessment of LLMs in Mental e-Health Applications

From 'Help' to Helpful: A Hierarchical Assessment of LLMs in Mental e-Health Applications

• Evaluated 11 LLMs generating six-word subject lines for German counselling emails. • Used hierarchical assessment: first categorize outputs, then rank within categories. • Nine a

Research & Labs · February 24, 2026 (updated February 24, 2026) · 1 min · 174 words
From Bias Mitigation to Bias Negotiation: Governing Identity and Sociocultural Reasoning in Generative AI

From Bias Mitigation to Bias Negotiation: Governing Identity and Sociocultural Reasoning in Generative AI

• LLMs leverage shared cultural patterns to interpret and act in social contexts. • Identity is central to judgment, yet current governance focuses solely on bias mitigation. • Bia

Research & Labs · February 24, 2026 (updated February 24, 2026) · 2 min · 225 words
Synthetic Media in Multilingual MOOCs: Deepfake Tutors, Pedagogical Effects, and Ethical-Policy Challenges

Synthetic Media in Multilingual MOOCs: Deepfake Tutors, Pedagogical Effects, and Ethical-Policy Challenges

• Deepfake and synthetic media are increasingly used as avatar tutors in multilingual MOOCs. • They boost social presence and participation by providing culturally relevant, multil

Research & Labs · February 24, 2026 (updated February 24, 2026) · 1 min · 183 words