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.
Yearly Review
Usually, 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.
Particularly for this week, we held a yearly review to observe how the Generative AI space changed and developed over this past year. We looked at how astonishingly quickly this field has evolved, and anticipate even more developments to arrive in 2025.
2024 AI and Technology Highlights
This year saw significant advancements across AI tools, regulations, and practical applications. Anthropic’s Claude models expanded globally, Google and OpenAI released groundbreaking updates like the o1 reasoning model and GPT-4 Turbo, while Meta and Microsoft continued innovating with the Llama and Phi series for on-device applications. Open-source AI thrived, with releases like Allen Institute’s OLMo and Hugging Face’s Common Models enhancing accessibility for researchers and developers.
Generative AI played a central role, driving education-focused initiatives such as UBC’s AI integration in classrooms and MIT’s resource hub for educators. Applications ranged from AlphaProteo’s breakthroughs in drug discovery to ChatGPT’s radiology-level tumor detection. Conversational AI matured, with real-time voice interactions from ElevenLabs and enhanced prompt tools like Anthropic’s library.
Legal and ethical challenges also surfaced, including copyright lawsuits over AI-generated art and debates about AI’s role in perpetuating colonialist patterns. Regulation took shape with the EU’s sweeping AI Act, and researchers tackled biases in LLMs while proposing solutions. AI even turned the tables on scammers, highlighting its growing role in cybersecurity.
Emerging innovations, like robotics for household chores and tools for decoding animal communication, reflected AI’s widening scope. The year ended with high expectations for protocols like Anthropic’s Model Context Protocol and advances in retrieval-augmented generation, setting the stage for even greater breakthroughs in 2025.
Here’s the overview of this year’s news:
January
Hugging Face grows its AI platform
Hugging Face continues to expand its ecosystem, offering advanced tools and models for developers to build and collaborate on machine learning projects. Read more.
NY Times and OpenAI face tensions over journalism and AI
OpenAI and The New York Times clash over the role of AI in journalism, particularly regarding data usage and content generation. Read more.
OpenAI launches the ‘GPT Store’
OpenAI introduces the GPT Store, enabling developers to create and share custom GPT-powered applications for various use cases. Read more.
Taylor Swift sparks tech regulation talks
Taylor Swift’s influence drives discussions on the regulation of deepfake content, highlighting broader technology and consumer protection issues. Read more.
February
Google expands AI model capabilities with long-context windows
Google announces advancements in AI models capable of processing significantly longer context windows, enhancing applications like summarization and research. Read more.
Google releases Gemini Small and open-source AI models
Google launches Gemini Small, a compact and efficient AI model, alongside open-source initiatives to foster broader adoption and collaboration. Read more.
Google addresses issues with Gemini image generation
Google investigates and resolves unintended outputs in its Gemini image-generation model, reinforcing its commitment to responsible AI development. Read more.
Lawyer faces fallout over AI misuse in court case
A legal case highlights challenges with AI-generated content, as a lawyer faces scrutiny for presenting fabricated data produced by AI tools. Read more.
March
Anthropic unveils Claude 3 models
Anthropic releases its Claude 3 AI family, offering improved safety, performance, and reasoning capabilities for advanced AI applications. Read more.
OpenAI previews Sora, a new AI system
OpenAI showcases Sora, an AI designed for natural language understanding, which officially launched nine months later. Read more.
Suno introduces V3 of its music generation model
Suno releases its third-generation music AI model capable of producing near human-level compositions and performances. Read more.
Semantic Scholar launches to support scientific research
Semantic Scholar provides a powerful tool for researchers, offering AI-driven assistance in finding and analyzing scientific literature. Read more.
EU passes comprehensive AI Act
The European Union approves the controversial AI Act, establishing broad regulations for AI use with mixed reactions from the tech community. Read more.
Cognition AI introduces Devin, later mired in controversy
Cognition AI’s Devin, a highly praised conversational agent, faced scrutiny over authenticity and was delayed until December. Read more.
CrewAI demonstrates real-world agent frameworks
CrewAI introduces practical applications of AI agent frameworks, showcasing their capabilities in handling complex tasks. Read more.
Hugging Face releases the Common Corpus for language modeling
Common Corpus, a dataset for large language model (LLM) training consisting entirely of data in the public domain, is released but sees limited usage for several months post-launch. Read more.
April
UBC updates Privacy Impact Assessments for classroom AI tools
UBC releases updated guidelines for integrating generative AI tools into classrooms while addressing privacy and ethical concerns. Read more.
ChatGPT becomes available without requiring an account
OpenAI makes ChatGPT accessible to users without an account, simplifying access to AI-powered chat capabilities. Read more.
OpenAI launches GPT-4 and GPT-4 Turbo models
OpenAI introduces GPT-4 and GPT-4 Turbo, offering faster response times and enhanced reasoning capabilities. Read more.
Anthropic enables tool usage with Claude 3 models
Anthropic announces tool usage integration for Claude 3, allowing AI models to interact with external tools for better functionality. Read more.
AI-native hardware experiments face challenges
Efforts to develop hardware optimized for AI hit setbacks, reflecting the complexity of creating AI-native devices. Read more.
Meta introduces Llama 3 for on-device AI applications
Llama 3 becomes Meta AI’s latest model, designed for effective use on personal devices with robust capabilities. Read more.
Microsoft launches Phi-3, a compact AI model for mobile devices
Microsoft unveils Phi-3, a lightweight and cost-efficient AI model designed to run effectively on smartphones and other portable devices. Read more.
May
AI reduces mortality rates in a Taiwanese hospital
AI technology implemented in two hospitals in Taiwan helped decrease all-purpose mortality by 17%, showcasing its potential in healthcare. Read more.
Nick Bostrom’s new book sparks interest in AI’s future
Renowned philosopher Nick Bostrom explores the challenges and opportunities presented by artificial intelligence in his latest publication. Read more.
OpenAI introduces ChatGPT-4o for broader use cases
ChatGPT-4o expands OpenAI’s capabilities, offering users enhanced, multimodal interactions and higher efficiency. Read more.
UBC’s Celebrate Learning Week highlights generative AI
UBC’s annual learning event focuses entirely on generative AI, exploring its applications and implications in education. Read more.
Socrates launched for AI-augmented dialogue
UBC releases Socrates, an AI-enhanced tool for facilitating Socratic dialogue to provide deeper insight into presuppositions and assumptions in argumentative discussions. Read more.
Microsoft’s ‘Recall’ feature sparks concerns over privacy
Microsoft’s Recall tool, designed to record device usage sessions, receives backlash for potential privacy risks and misuse. Read more.
OpenAI disbands its safety-focused team
OpenAI dissolves its safety team, raising concerns about its commitment to addressing risks associated with advanced AI. Read more.
Kenyan tech workers pen an open letter on labor rights
Kenyan tech workers urge President Biden to address labor exploitation in the global tech industry, advocating for fair treatment. Read more.
June
Claude chatbot launches in Canada, becoming a popular choice
Anthropic’s safety-focused Claude chatbot expands to Canada, gaining widespread adoption due to its usability and reliability. Read more.
Anthropic introduces a prompt library for AI interactions
The prompt library helps users craft better AI inputs, with hints suggesting that Claude may also refine poorly written prompts automatically. Read more.
AI works to decode dog vocalizations
Researchers explore using AI to interpret dog sounds, potentially enabling better communication between humans and their pets. Read more.
AI studies explore elephant communication patterns
Scientists use AI to study elephants’ vocalizations, uncovering insights about their complex social interactions. Read more.
Research highlights biases in prompt personas with solutions proposed
A study reveals that AI-generated personas in prompts may amplify biases but offers potential methods to mitigate these issues. Read more.
UBC launches large TLEF funding call for generative AI initiatives
UBC announces a special grant program encouraging collaborative projects to integrate generative AI into academic environments. Read more.
Paper critiques misinformation in AI, offering paths forward
A thought-provoking paper examines how AI systems spread misinformation, advocating for greater accountability in their design and usage. Read more.
July
Kyutai enables low-latency audio-to-audio AI conversations
Kyutai demonstrates the feasibility of real-time audio-to-audio AI communication, opening possibilities for smoother conversational AI experiences. Read more.
WebSim provides a simulated browser for AI experiments
WebSim introduces an interactive platform using AI to create and interact with web environments in a controlled simulation. Read more.
Stanford launches STORM for research knowledge curation
STORM, Stanford’s new AI-driven tool, assists researchers in organizing and extracting insights from large datasets and scientific knowledge. Read more.
OpenAI releases GPT-4o Mini for cost-effective intelligence
GPT-4o Mini offers comparable performance to larger models at significantly lower costs, making advanced AI more accessible. Read more.
Meta introduces Llama 3.1 for versatile AI applications
Llama 3.1 builds on Meta’s Llama series, offering improvements in scalability and efficiency for AI research and deployment. Read more.
Mistral releases its Large 2 model with advanced capabilities
Mistral’s Large 2 model debuts, providing significant enhancements for tasks requiring high-performance AI solutions. Read more.
AI aids in identifying additional forms of Parkinson’s disease
AI tools assist researchers in categorizing Parkinson’s subtypes, leading to more tailored diagnosis and treatment options. Read more.
AI achieves silver medal-level performance in math competitions
DeepMind’s AI solves International Math Olympiad problems at a competitive level, showcasing advancements in problem-solving abilities. Read more.
August
Flux1 introduces a new open-source text-to-image model
Based Labs releases Flux1, a competitive open-source diffusion model for text-to-image generation. Read more.
Glif app revives workflow automation akin to Yahoo Pipes
Glif offers a powerful platform for creating and customizing workflows, reminiscent of the popular Yahoo Pipes tool. Read more.
AI-powered robots showcase advanced capabilities
Figure Robotics demonstrates robots capable of performing complex tasks, hinting at a future of automation in everyday life. Read more.
OpenAI introduces structured outputs for API responses
Structured outputs allow developers to specify response formats for OpenAI models, ensuring more precise and usable outputs, especially for coding purposes. Read more.
AI-generated art faces legal challenges in copyright lawsuits
Stability AI and MidJourney are sued by artists, raising questions about intellectual property rights for AI-generated content. Read more.
Judicial redaction errors highlight privacy concerns
High-profile cases of poor redaction practices serve as reminders of the risks of mishandling sensitive data. Read more.
September
MIT launches AI Resource Hub for educators
MIT introduces an AI Resource Hub to support educators in effectively integrating AI into their teaching practices. Read more.
Robots showcase advanced real-world applications
A new demonstration for NEO Beta highlights the growing capabilities of robotics, with applications for various household tasks. Read more.
AI aids in drug discovery and disease research
DeepMind’s AlphaProteo generates novel proteins, advancing the understanding of diseases and accelerating drug development. Read more.
OpenAI previews o1, a reasoning-focused AI model
OpenAI’s o1 model demonstrates enhanced reasoning capabilities, solving complex scientific problems at a high level. Read more.
OpenAI’s o1 solves PhD-level physics problems
The o1 model successfully tackles PhD level advanced physics code, showcasing its potential for academic and research applications. Read more.
UBC develops an AI course for K-12 educators
UBC’s Faculty of Education launches a course to explore opportunities and challenges in integrating AI into K-12 classrooms. Read more.
ChatGPT gets Advanced Voice Mode for real-time interaction
OpenAI introduces Advanced Voice Mode, enabling near-instantaneous voice-based conversations with ChatGPT. Read more.
Meta releases Llama 3.2 for edge and mobile applications
Llama 3.2 improves on Meta’s previous models with optimized performance for mobile and edge devices. Read more.
Google introduces Notebook LM for document analysis
Notebook LM leverages AI to help users better organize and extract insights from their documents. Read more.
October
ChatGPT rivals radiologists in brain tumor detection
ChatGPT demonstrates capabilities in brain tumor recognition, achieving results comparable to trained radiologists in a clinical study. Read more.
Paper explores how LLMs can integrate external data effectively
A new study highlights strategies for enabling large language models to use external data, improving their accuracy and adaptability. Read more.
Geoffrey Hinton and John Hopfield wins Nobel Prize for AI contributions
Geoffrey Hinton and John Hopfield, pioneers in AI, are awarded the Nobel Prize in Physics for their groundbreaking work in machine learning. Read more.
Debate grows over whether AI represents a new form of colonialism
A thought-provoking article examines the potential of AI technologies to reinforce historical patterns of economic and cultural dominance. Read more.
Google launches Illuminate, an AI podcast for document exploration
Illuminate uses AI to generate podcasts summarizing and discussing the content of uploaded documents for easier comprehension. Read more.
Meta open-sources its AI podcast generator tool
Meta releases an open-source version of Google’s AI podcast generator, enabling developers to create similar applications. Read more.
OpenAI unveils an educational framework for agent orchestration
OpenAI introduces “Swarm,” a framework designed to educate and enable the use of AI agents working collaboratively on tasks. Read more.
November
Robots may soon handle mundane household chores
Advancements in robotics bring closer the possibility of AI-powered robots performing everyday tasks like cleaning and cooking. Read more.
AI shows promise in understanding humans better than we can
AI systems are argued to demonstrate the potential to decode complex human emotions and behaviors with higher accuracy than humans in some scenarios. Read more.
AI predicts human behavior with remarkable precision
Research indicates AI’s increasing ability to forecast human actions, outperforming traditional methods in accuracy and efficiency. Read more.
Ethical public sector AI usage explored
JK-M proposes guidelines for the ethical use of AI in government operations, emphasizing transparency, fairness, and accountability. Read more.
MIT releases Undermind, a tool for scientific discovery
Undermind is launched as an AI-powered research assistant to help scientists uncover hidden insights in large datasets. Read more.
December
Scammers fall victim to AI-powered countermeasures
AI technology named ‘Daisy’ turns the tables on scammers, highlighting innovative uses of AI to combat fraud. Read more.
Anthropic unveils the Model Context Protocol
The Model Context Protocol introduces a new standard for managing contextual information in AI systems, promising significant advancements in the coming year. Read more.
Allen Institute releases an open-source AI model
The Allen Institute launches OLMo, a truly open-source AI model, aiming to provide transparency and accessibility for researchers. Read more.
Hugging Face releases ‘Common Models’ for public use
Hugging Face expands its open-source ecosystem with Common Models, notable for being trained on exclusively public data, offering accessible tools for AI experimentation and development. Read more.
Important research published on retrieval-augmented generation (RAG)
A new paper advances understanding of RAG techniques, improving how AI systems retrieve and incorporate external information into responses. Read more.
Conversational AI agents gain voice capabilities
ElevenLabs debuts conversational AI agents with real-time voice capabilities, enabling more natural human-AI interactions. Read more.
So how far have we come?
Here is a visual comparison of Will Smith eating spaghetti from Late 2023 compared to Mid 2024.
Late 2023, you really needed top-tier hardware to run LLMs on your own device, easily in the thousands of dollars per person using them.
And today (published the day of the Studio in West Coast Time), a $249 (USD) supercomputer was released that can get 30+ tokens per second running some of the tougher local language models. Full article here.
Tool of the Week (Returning next session!)
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 […]