GenAI Studio: News, Tools, and Teaching & Learning FAQs
These sixty minute, weekly sessions – facilitated by Technologists and Pedagogy Experts from the CTLT – are designed for faculty and staff at UBC who are using, or thinking about using, Generative AI tools as part of their teaching, researching, or daily work. Each week we discuss the news of the week, highlight a specific tool for use within teaching and learning, and then hold a question and answer session for attendees.
They run on Zoom every Wednesday from 1pm – 2pm and you can register for upcoming events on the CTLT Events Website.
News of the Week
Each week we discuss several new items that happened in the Generative AI space over the past 7 days. There’s usually a flood of new AI-adjacent news every week – as this industry is moving so fast – so we highlight news articles which are relevant to the UBC community.
This week in AI, new model releases and applied innovations spanned healthcare, education, reasoning, and interpretability. A non-invasive model claims 99% accuracy in brain tumour diagnosis, though questions remain about clinical readiness. Anthropic advanced Claude’s role in education and app development, while Microsoft and Moonshot AI introduced compact reasoning and sparse trillion-parameter models, respectively. Mistral launched Voxtral for open-source audio analysis, and xAI debuted Grok 4 with real-time search, drawing both interest and skepticism. Meanwhile, researchers debated chain-of-thought traceability as a fragile but vital safety tool in AI oversight.
Here’s this week’s news:
AI Diagnosing Brain Tumors Without Surgery
A novel AI model achieves an impressive 99% accuracy in diagnosing brain tumors by analyzing non-invasive imaging techniques, potentially reducing the need for surgical biopsies. The model leverages deep learning on MRI datasets to distinguish tumor types rapidly and accurately. This breakthrough could revolutionize patient care by lowering risk, cost, and wait times.
Anthropic Advances Claude for Education
Anthropic has rolled out integrations of Claude with Canvas, Panopto, and Wiley to enrich educational environments. The updated Claude for Education enables students to engage with course materials—including readings and lecture media—directly within the AI interface. This evolution supports deeper student engagement and academic integrity.
Microsoft releases new model Phi‑4 Mini Flash‑Reasoning
Microsoft’s Phi‑4‑mini‑flash‑reasoning is a distilled 3.8B‑parameter model optimized for multi‑step mathematical and logical reasoning under tight resource and latency demands. Released on Hugging Face, it excels in formal proofs, symbolic math, and structured analysis. It offers a balance of efficiency and capability for edge and mobile deployments.
Try Phi‑4‑mini‑flash‑reasoning yourself.
Voxtral: Mistral’s Open‑Source Speech Understanding Suite
Mistral AI has released Voxtral, an open‑source family of speech‑understanding models designed for transcription, multilingual comprehension, summarization, and function‑triggering from spoken input. Early evaluations indicate it outperforms Whisper on multilingual tasks, marking a milestone in open voice-tech.
Kimi‑K2: Moonshot AI’s 1T‑Parameter MoE Model
Benchmarks indicate that Kimi‑K2, Moonshot AI’s mixture‑of‑experts model with 1 trillion parameters (32B active), matches or exceeds leading Western models like GPT‑4 in coding and reasoning tasks. The open‑source release allows researchers and developers to run it locally or via agents.
Grok 4: xAI’s Real‑Time, Musk‑Influenced Model
xAI has released Grok 4 (and Grok 4 Heavy) with a tiered subscription model. It integrates real-time web/X search, tool use, and multi-agent reasoning, outperforming peers on benchmarks like ARC‑AGI and Humanity’s Last Exam. However, experts have flagged that it sometimes searches Elon Musk’s X posts when answering controversial questions, raising concerns about bias. Grok 4’s release is overshadowed by prior antisemitic content, spotlighting challenges in AI content safety despite technical progress.
View Simon Willison’s analysis of Musk search behaviour.
Chain‑of‑Thought Monitoring: A Fragile Safety Avenue
Tomek Korbak et al. argue that monitoring AI “chain‑of‑thought” (CoT) offers a unique route for safety oversight since the model’s internal reasoning is exposed in human‑readable form. However, this method is fragile—susceptible to architectural changes and adversarial prompts—which may undermine long-term efficacy. They advocate ongoing investment in CoT monitoring alongside complementary safety techniques.
Read the Tomek Korbak et al.’s paper (PDF).
Tool of The Week: Claude Artifacts – AI Powered Project Scaffolding
What are Claude Build Artifacts?
Claude Artifacts is a new feature from Anthropic that allows users to create and edit persistent outputs—such as documents, code, diagrams, or applications—directly within the Claude interface. Unlike typical conversational exchanges that disappear, Artifacts remain visible alongside the chat, enabling an ongoing, editable workspace. This innovation bridges the gap between chat-based interaction and real-time content development.
How is it used?
Artifacts are generated through dialogue with Claude, where the user prompts the model to create something—such as a website, flashcard app, or structured document—and the resulting content appears in a dedicated side panel. Users can iteratively update or refine the Artifact through continued conversation, enabling a collaborative workflow that blends text input with visual editing. This makes it well suited for structured, multi-step creation processes.
What is it used for?
Claude Artifacts is designed for tasks that benefit from persistent, editable content—such as software prototyping, writing, lesson planning, game creation, and interactive design. It shifts the AI experience from ephemeral chat to durable output, allowing users to build, revise, and explore ideas in a more structured and visible environment. This reflects a broader trend of LLMs evolving from assistants into content development platforms.
Check out the release information.
Try Claude Artifacts out yourself.
Questions and Answers
Each studio ends with a question and answer session whereby attendees can ask questions of the pedagogy experts and technologists who facilitate the sessions. We have published a full FAQ section on this site. If you have other questions about GenAI usage, please get in touch.
-
Assessment Design using Generative AI
Generative AI is reshaping assessment design, requiring faculty to adapt assignments to maintain academic integrity. The GENAI Assessment Scale guides AI use in coursework, from study aids to full collaboration, helping educators create assessments that balance AI integration with skill development, fostering critical thinking and fairness in learning.
-
How can I use GenAI in my course?
In education, the integration of GenAI offers a multitude of applications within your courses. Presented is a detailed table categorizing various use cases, outlining the specific roles they play, their pedagogical benefits, and potential risks associated with their implementation. A Complete Breakdown of each use case and the original image can be found here. At […]