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.
Here’s this week’s news:
ZDNet on GPT-4 Turbo vs. Claude 3 – GPT-4 Turbo has regained the title of the best AI model, surpassing Anthropic’s Claude 3, in a recent evaluation that considers performance, versatility, and technological advancements.
BBC Article on Humane AI Pin – The AI Pin, developed by Humane, has been criticized for its practical performance despite its innovative design. The device, intended to enhance daily interactions without a screen, struggled with functionality issues, poor visibility in bright light, rapid overheating, and frequent battery recharges. Despite the backlash, Humane acknowledges the feedback and plans improvements, reflecting the ongoing challenge of integrating AI into consumer technology.
Udio – Make Your Music – Udio offers a platform where users can create, collaborate, and engage with music using AI-driven tools. It aims to simplify music production and enhance creativity, making it more accessible to enthusiasts and professionals alike by integrating innovative technology into the music creation process.
AI-Animated Pink Floyd Music Video Stirs Debate The release of an AI-animated music video for a Pink Floyd song has sparked discussions across the music and tech industries. The debate centers on the creative implications and the authenticity of using AI in producing music visuals.
Hugging Face Introduces Idefics2 – Hugging Face has launched Idefics2, an 8B parameter vision-language model that significantly improves upon its predecessor, Idefics1, in terms of OCR capabilities and multimodal integration. The model is designed to handle a range of tasks like visual question answering and text generation from images, and is accessible for community use under an open license.
Buildbox 4 AI Game Development – Buildbox has released its latest game development software, Buildbox 4, which incorporates AI tools to streamline game design and development processes, making it easier for developers to create complex games.
Concerns Over AI’s Impact on LGBTQ Community – There are growing concerns about how AI technologies might impact the LGBTQ community, particularly regarding biases in AI algorithms and how they could reinforce harmful stereotypes or discriminate against marginalized groups.
Study on AI-Driven Persuasion in Debates – A research explores the persuasive capabilities of large language models (LLMs) in structured debates involving humans, comparing scenarios with and without access to personal data. Results indicate that personalization markedly increases LLMs’ persuasiveness, underscoring implications for AI regulation and design in digital platforms.
Tool of the Week
Each week we demonstrate a Generative AI tool that can be used within teaching and learning. The GenAI space is evolving rapidly, and as such we demo new tools or new ways people use those tools.
As a reminder not all tools we showcase have successfully been through the PIA process at UBC.
This week’s Tool of the Week: Enchanted
What is Enchanted?
Enchanted is a specialized chat application tailored for researchers and developers working with large language models (LLMs). It provides a unique platform for interacting with self-hosted models, facilitating real-time conversations and testing. The application integrates with the Ollama API and supports all models within this ecosystem, making it a versatile tool for those deeply involved in LLM research and development.
How is it used?
To use Enchanted, users must have an operational Ollama server, as the application requires specifying the server endpoint in its settings. This setup allows users to leverage the advanced features of Enchanted, including streaming capabilities and the latest Chat API that supports conversation context. The interface supports Markdown formatting, enabling the display of tables, lists, and code in a structured and visually appealing manner. Users can seamlessly switch between different models to test various prompts, making it an efficient tool for model evaluation and interaction.
What is it used for?
Enchanted is designed primarily for the LLM community, particularly those who manage and test self-hosted models. Its capabilities are crucial for developing, refining, and deploying language models, as it allows researchers to interact directly with their models in a controlled environment. The tool’s support for multiple model integrations and its advanced communication features make it indispensable for experimental setups and for researchers aiming to push the boundaries of what LLMs can achieve. Though still in the early stages of development, Enchanted’s robust roadmap indicates its potential to become a leading tool in the LLM research field.
Sample Prompt For Grading a Short Answer Question
You are an experienced professor of computer science and are teaching a 200-level software engineering course at a large University in Canada. You are tasked with grading student answers to the following question: “How do you get multiple inheritance in Java? Answer in 3 sentences or fewer.”. The maximum a student can score for this question is 5 marks. A great answer scores 5 marks, an average answer scores 3 marks, and an answer that doesn’t directly answer the question scores 0 marks. Here are example answers at various scoring levels, with reasoning for each score:
Score: 5/5 Points.
Answer: “Java avoids multiple inheritance with classes to prevent complexity like the diamond problem, but allows a class to implement multiple interfaces. This approach lets a class inherit multiple sets of method definitions without the issues associated with traditional multiple inheritance. Thus, multiple inheritance in Java is facilitated through interfaces rather than classes.”
Reasoning: This answer clearly explains the concept of multiple inheritance in Java, mentioning the specific problem that Java avoids by not allowing class-based multiple inheritance, and how interfaces solve this issue, covering both the theoretical and practical aspects comprehensively.
Score: 4/5 Points.
Answer: “Multiple inheritance of state is not possible in Java with classes to prevent ambiguity. Instead, Java supports multiple inheritance of behavior through interfaces, allowing a class to implement more than one interface.”
Reasoning: The response is accurate and explains the key concepts, though it could further enhance its score by detailing a little more on how interfaces allow inheritance of behaviors or giving a brief explanation of the kind of problems (like the diamond problem) that Java aims to avoid.
Score: 3/5 Points.
Answer: “Since Java classes cannot inherit from more than one class, Java uses interfaces to allow classes to inherit from multiple sources.”
Reasoning: This answer correctly identifies the mechanism Java uses to simulate multiple inheritance but lacks detail on why Java restricts multiple class inheritance and what inheriting from multiple interfaces entails.
Score: 2/5 Points.
Answer: “To handle multiple inheritance, Java allows implementing multiple interfaces as classes cannot have more than one superclass.”
Reasoning: This answer points out the basic structure of Java’s approach to multiple inheritance but fails to explain why this limitation exists or how interfaces offer a solution beyond merely being an available option.
Score: 1/5 Points.
Answer: “Java does not allow multiple class inheritance but uses interfaces instead.”
Reasoning: This statement, while technically correct, is overly simplistic and lacks depth. It fails to provide any explanation of how interfaces serve as a functional alternative to multiple inheritance or why Java chooses this path.
Score: 0/5 Points.
Answer: “You can inherit from multiple classes in Java by declaring more than one superclass.”
Reasoning: This response is fundamentally incorrect as it suggests a possibility that does not exist in Java, demonstrating a lack of understanding of Java’s inheritance model.
Using that marking scheme, now please provide a score – out of 5 points – for the following answer by a student:
“Just extend multiple classes in Java to get multiple inheritance.”
Only provide the score out of 5. i.e. your answer should be only “X/5” where X is the score you suggest for this answer. It is VITAL that you follow the marking scheme above when considering the student’s answer. Do not provide any reasoning or explanation, just the score out of five.
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 […]