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.
In this week’s tech news, Amazon introduced Nova, a suite of generative AI models designed to create text and images, strengthening its position in the AI sector. AWS also launched a service to address AI hallucinations, enhancing the reliability of AI-generated outputs. OpenAI clarified that ChatGPT’s failure to recognize ‘David Mayer’ was caused by a technical glitch rather than data limitations. Canadian news publishers filed a lawsuit against OpenAI, alleging unauthorized use of their content to train ChatGPT. MLCommons released AILuminate, a benchmark assessing AI models’ safety by evaluating responses across various hazard categories. DeepMind announced Genie 2, a large-scale world model designed to advance AI understanding of complex environments. Meta reported that generative AI had little impact on global elections this year, with limited spread of AI-generated misinformation. Research explored optimal retrieval methods for retrieval-augmented generation (RAG) systems, improving question-answering performance. A recently published book on generative AI and education discussed the evolution of generative AI and its application in the field of education. The Centre for Teaching, Learning and Technology (CTLT) at UBC is holding a workshop on integrating generative AI into assignments and assessments. Finally, Google introduced an experimental chess platform, GenChess, that allows users to create their own unique chess set design to play chess.
Here’s this week’s news:
Amazon Unveils Nova AI Models
Amazon has introduced Nova, a suite of generative AI models designed to create text, images, and videos, aiming to strengthen its position in the AI sector. These models, available through Amazon Bedrock, offer varying levels of complexity to assist businesses in building generative AI applications. The Nova lineup includes:
- Amazon Nova Micro: A text-only model focused on low-latency and low cost, optimized for fast responses.
- Amazon Nova Lite: A cost-effective multimodal model optimized for speed, capable of processing images, videos, and text to generate text outputs.
- Amazon Nova Pro: A highly capable multimodal model offering a balance of accuracy, speed, and cost for a wide range of tasks.
- Amazon Nova Premier: Coming soon, this model is expected to be Amazon’s most capable multimodal model for complex reasoning tasks, with availability planned for early 2025.
- Amazon Nova Canvas: An image generation model that accepts text and image inputs to produce high-quality images, designed to deliver easily customizable visual content.
- Amazon Nova Reel: A video generation model that accepts text and image inputs to produce high-quality videos, offering control over visual content generation.
Additionally, Amazon plans to release a speech-to-speech model and a native multimodal-to-multimodal model later in 2025. These developments signify Amazon’s attempt to catch up in the AI race and cater to industries like entertainment.
Further explore the Amazon Nova models.
AWS Introduces Automated Reasoning checks to Combat AI Hallucinations
Amazon Web Services (AWS) has launched Automated Reasoning checks within its Bedrock service to address AI hallucinations—instances where AI models generate incorrect or misleading information. This tool cross-references model outputs with customer-provided data to ensure accuracy, aiming to enhance the reliability of AI-generated responses. Rather than attempting to eliminate hallucination, this approach seeks to add a safeguard so that hallucinated outputs can be identified and avoided.
OpenAI Addresses ChatGPT ‘David Mayer’ Glitch
OpenAI has clarified that ChatGPT’s inability to generate the name “David Mayer” was due to a system glitch that mistakenly flagged the name, preventing its appearance in responses. Many speculations have been brought up regarding the potential underlying reason for this glitch, but OpenAI has declined to comment further. A fix has been issued for the “David Mayer” case in particular, but other names are reported to be still triggering a similar error when entered into ChatGPT.
Canadian News Publishers Sue OpenAI
A coalition of Canadian news publishers, including The Canadian Press, Torstar, The Globe and Mail, Postmedia, and CBC/Radio-Canada, has filed a lawsuit against OpenAI, alleging that the company used their content without permission to train its ChatGPT AI system. The publishers claim that OpenAI’s practices undermine their investments in journalism and violate copyright laws. OpenAI asserts that its models are trained on publicly available data and therefore adhere to fair use principles. This marks the first such legal action in Canada, though similar cases have been initiated in the United States.
MLCommons Launches AILuminate Benchmark
MLCommons has released AILuminate, a benchmark designed to assess the safety of AI systems by evaluating large language models on their responses to over 24,000 test prompts across various categories divided between physical, non-physical, and contextual hazards. The AILuminate Benchmark is designed to be applied to AI models to “produce ‘safety grades’ for general purpose chat systems”. This initiative seeks to standardize AI safety evaluations and promote the responsible development and deployment of AI systems.
Read the AILuminate Benchmark overview, and see the results of leading AI models on this benchmark.
Google DeepMind Introduces Genie 2
Google DeepMind has unveiled Genie 2, an advanced AI world model capable of generating interactive 3D environments from text prompts or images. Building upon its predecessor, Genie 2 enhances interactivity with physics-based actions, allowing users and AI agents to explore dynamic virtual worlds featuring realistic lighting effects and object manipulation. Not only applicable to users, this model is aimed at providing a “limitless curriculum of novel worlds” for AI agents, acting as a training environment for AI development. This innovation holds significant potential for applications in gaming, digital art, and AI agent training, enabling the rapid prototyping of interactive experiences.
Explore Google Deepmind’s Genie 2.
Meta Reports Minimal AI Impact on 2024 Elections
Despite widespread concerns that generative AI could interfere with major elections worldwide in 2024, Meta Platforms reported that the technology had a limited impact across its applications. They reported that volumes of false content generated by AI were low, and Meta was able to promptly handle the issue by labelling or removing the content, leaving accounts attempting to share AI-generated misinformation failing to gather a significant audience.
Research on Retrieval-Augmented Generation (RAG)
A recent study titled “Toward Optimal Search and Retrieval for RAG” examines how retrieval components affect the performance of retrieval-augmented generation (RAG) systems, particularly in question answering tasks. The research indicates that reducing search accuracy has minimal impact on RAG outcomes while potentially enhancing retrieval speed and memory efficiency. These findings offer valuable insights for developing high-performance RAG pipelines.
Generative AI and Education: Digital Pedagogies, Teaching Innovation and Learning Design
A recent book written by B. Mairéad Pratschke discussed the context of generative AI in the field of Education, discussing how the landscape is shifting around the use and application of generative AI. Topics extend from providing historical and technological context to the current state of generative AI in education, to future implications of integrating AI into educational contexts. This book is recommended to those interested in understanding the evolution of generative AI in the educational context.
Read the book through the UBC Library.
Integrating GenAI in Assignments and Assessments – December 17, 2024
The University of British Columbia’s Centre for Teaching, Learning and Technology (CTLT) is hosting an online workshop exploring various methods of incorporating GenAI into academic assignments and assessments. Topics include prohibiting GenAI use, leveraging it for ideation and brainstorming, evaluating GenAI outputs, and enabling students to use GenAI as a co-pilot. The workshop will also present examples from university teaching contexts across different disciplines. While the general theme remains consistent, new topics and examples are presented each session, so continuing registrants are encouraged to attend.
Google Introduces GenChess: AI-Driven Customizable Chess Experience
Google has launched GenChess, an experimental AI-powered chess platform that allows users to create custom chess pieces by inputting short descriptions, with the AI generating themed sets for both players. The game utilizes Google’s Imagen 3 AI model to produce these unique designs, offering a creative twist on traditional chess. Players can choose from three difficulty levels and two time controls, and can switch between isometric and top-down views for gameplay. This release complements the announcement of a forthcoming chess bot within Google’s AI chatbot, Gemini, expected to be available to Gemini Advanced subscribers in December.
Tool of the Week
Tool of the Week: ElevenLabs’ Conversational AI
What is ElevenLabs’ Conversational AI?
ElevenLabs’ Conversational AI is a platform that allows developers to create voice-interactive agents capable of engaging in natural, real-time conversations across various applications.
How is it used?
Developers can integrate this tool into web or mobile applications, enabling agents to process speech input, generate contextually appropriate responses, and handle dynamic interactions seamlessly.
What is it used for?
This platform is utilized in sectors like customer support, education, and gaming to enhance user engagement through interactive voice agents, offering personalized experiences and efficient information delivery.
For additional information, explore ElevenLabs’ Conversational AI.
Without a PIA, instructors cannot require students use the tool or service without providing alternatives that do not require use of student private information
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 […]