💬 Reasoning Models Are Now Standard, So Expect Better Responses
Paid subscribers to Google Gemini now have access to a “thinking” model, Gemini 2.5 Pro, built on Google’s first thinking model, “Flash”. Thinking models aim for more nuance and context in their responses.
Learn More: Gemini 2.5: Our Most Intelligent AI Model
✍️ 10 “Exemplary” University AI Policies
A handy resource as the Provost convenes the campus AI Summit to discuss what should be next for MSU. The list is at the end of the article.
Learn More: Crafting Thoughtful AI Policy in Higher Education: A Guide for Institutional Leaders
🔎 AI Detection is Domain Specific
Fine tuning AI detection models work better when customized for domain-specific context. The BERT model jumped from 53% detection accuracy to 97% when finetuned with relevant academic text.
Learn More: Campino, J. Unleashing the transformers: NLP models detect AI writing in education. J. Comput. Educ. (2024).
🤝 Adult Learners Trust Humans More Than AI
Engineering students who used ChatGPT with guidance showed stronger critical thinking and information literacy, but only when the instructor redesigned the assignment to include AI use, search strategy, and iterative research as graded components.