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 and education, the ETUG Spring 2025 Workshop highlights a shift back to tactile, hands-on approaches in educational technology, emphasizing how instructors practically implement edtech solutions in real-world settings. At UBC, the POSE Open Research Chat explores how generative AI reshapes perceptions of critical thinking in academic contexts; while a related event delves into the intersection of GenAI and Open Educational Resources, examining how these tools can enhance or complicate teaching and learning. In AI development news, DeepMind unveils AlphaEvolve, a Gemini-powered coding agent that autonomously discovers and optimizes complex algorithms, demonstrating results at or better than the leading solutions. FAL AI introduces F-Lite, a 10B parameter diffusion model designed to generate copyright-safe images, paired with a public demo showcasing its high-quality, safe text-to-image capabilities. On the prompt engineering front, Upsun’s guide outlines advanced prompting techniques to enhance coding outputs from language models, providing a framework for more structured and effective generation strategies. Anthropic expands its Claude assistant with a new Web Search API, enabling more informed and context-aware answers by incorporating up-to-date web results into responses. The Kokoro project debuts as a lightweight text-to-speech model with just 82M parameters, providing efficient audio generation suitable for resource-constrained environments. Finally we cover the pricing breakdown of the UBC GenAI toolkit, including the baseline costs, surge pricing, and plans for current and future users.
Here’s this week’s news:
ETUG Spring 2025 Workshop
Scheduled for May 22–23 at UBC Vancouver, the ETUG Spring 2025 Workshop, titled “Show Me How You Did It: Getting Back to Hands-On in EdTech,” emphasizes practical engagement with educational technologies. Participants will explore tools like VR, AR, and 3D capture, and engage in discussions on digital learning strategies and trends. The event fosters peer-led learning and collaboration among faculty, instructional designers, and technologists, creating an environment that encourages experimentation and showcases real-world implementations of emerging tools.
Explore the event details and register.
POSE Open Research Chat on Generative AI and Critical Thinking
On June 23, this online session will delve into how Generative AI (GenAI) affects perceptions of critical thinking skills in educational and professional contexts. Based on a 2025 study by Hank Lee and colleagues, the discussion will focus on self-reported reductions in cognitive effort and confidence among knowledge workers who rely heavily on GenAI tools. The event will prompt participants to reflect on the implications of AI-assisted reasoning and explore how open research practices can adapt to maintain critical inquiry.
Explore the event details and register.
GenAI and Open Educational Resources in Teaching and Learning
Scheduled for July 16, this interactive POSE session explores the integration of GenAI with Open Educational Resources (OER) in post-secondary education. Participants will examine how generative models can support the creation of dynamic, accessible, and interactive materials, such as AI-assisted textbooks. In addition to hands-on activities like collaboratively generating a chapter, the session addresses legal and ethical considerations including copyright, open licensing, and data privacy.
Explore the event details and register.
AlphaEvolve: DeepMind’s AI Agent for Algorithm Innovation
DeepMind has introduced AlphaEvolve, an AI system that surpasses human capabilities in designing certain algorithms. By combining Gemini AI’s coding abilities with testing and evolutionary techniques, AlphaEvolve has created algorithms more efficient than long-standing human-devised methods, including surpassing the 56-year-old Strassen algorithm for matrix computations. The system has also optimized solutions for practical tasks like data center scheduling and chip design.
Read the full article covering AlphaEvolve’s results and algorithm innovation.
F Lite: A Copyright-Safe Diffusion Model for Text-to-Image Generation
F Lite is a light-weight 10-billion-parameter diffusion model developed by Freepik in collaboration with Fal, designed specifically for text-to-image generation using only copyright-safe and SFW (safe for work) training data. Unlike many competing models, F Lite prioritizes legal clarity and ethical standards by excluding copyrighted or NSFW content from its dataset, making it suitable for professional and educational contexts. The model provides both API access and a web interface, allowing developers and creatives to generate high-quality visuals from text prompts with ease. Its streamlined deployment and licensing make it particularly attractive for organizations requiring compliant generative image tools.
Explore the F Lite Codebase on GitHub
Advanced Prompting Techniques for Coding
This article from Upsun DevCenter outlines sophisticated strategies for crafting more effective prompts when working with AI-assisted coding tools. It emphasizes the importance of structuring prompts with clarity and context, guiding the AI toward more accurate, efficient, and maintainable code output. Techniques include refining system-level instructions, anticipating token-level behaviors, and chaining prompts to build complex logic step by step. The piece serves as a practical guide for developers looking to optimize their interactions with large language models in real-world coding workflows.
Read more on Advanced Prompting Techniques for Coding.
Anthropic’s Claude Gains Web Search Capabilities
Anthropic has added real-time web search functionality to Claude, its AI assistant, enabling it to retrieve current information from the internet for enhanced responses. Initially available to paid users in the U.S., this feature allows Claude to cite sources and provide verification for its outputs, improving transparency and relevance. This development reflects a broader trend toward hybrid AI models that integrate static training data with live search to stay up-to-date. Anthropic plans to expand access to free-tier users and additional regions in the near future.
Explore the new feature and Anthropic’s future plans for integration.
Kokoro: A Lightweight Text-to-Speech Model
Kokoro is an open-source, lightweight text-to-speech (TTS) model optimized for real-time voice synthesis on consumer-grade hardware. Developed with efficiency in mind, Kokoro supports ONNX Runtime and includes a Gradio demo for easy experimentation and deployment. The repository offers a straightforward interface for integrating natural-sounding voice generation into web applications, mobile tools, or embedded systems. Its minimal footprint and flexible licensing make it a useful alternative to larger, resource-intensive TTS models.
Explore the Kokoro repository on GitHub.
UBC AI Toolkit Pricing and Availability Breakdown
This flow chart illustrates how UBC’s AI infrastructure enables scalable, centralized AI access for a growing number of TLEF-funded teaching and learning applications. Through a single load balancer and key management layer, requests are dynamically routed to a mix of always-on baseline compute and flexible, on-demand instances capable of absorbing surges in usage. The architecture is designed for future growth, allowing new projects to plug in easily as demand for AI-enhanced learning tools continues to expand.
Tool of The Week: Zed, an AI IDE Supporting Local Model Integration
What is Zed?
Zed is a fast, open-source code editor built in Rust, designed for high performance, collaboration, and AI integration. Created by the developers behind Atom and Tree-sitter, it offers a streamlined, GPU-accelerated environment that supports both solo and team-based development. What distinguishes Zed is its native support for integrating AI tools directly into the editor, including the ability to run your own local large language models (LLMs) for code assistance.
How is it used?
Zed is used as a full-featured development environment where users can write, navigate, and refactor code with real-time feedback and support. Developers can collaborate in a multiplayer setting, share sessions instantly, and invoke AI features without relying on external cloud services. By enabling local LLM hosting, Zed gives users the flexibility to work offline or with custom models tailored to their projects and privacy requirements.
What is it used for?
Zed is primarily used for writing and managing code across a wide range of programming languages and project types. It is especially valuable for developers who want an AI-augmented workflow with minimal latency and more control over data by running models locally. Whether for solo development, pair programming, or team collaboration, Zed enhances productivity with tightly integrated, customizable AI tools built into the editing experience.
Learn more about how to use Zed.
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 […]