Generative AI Literacy

The AI Universe: a graphic with nested circles showing the different types of AI tools. The smallest circle at the centre/bottom of the graphic is Generative AI, which involves language modeling, transfer learning, transformer architecture, self-attention menchanism, natural language understanding, generation, summarization, and dialogue systems.

The second circle, expanding out and holding the generative AI circle is labelled Deep learning, and includes deep neural networks (DNNs), transfer leraning, generative adversarial networks (GANs), Deep belief networks (DBNs), deep conovultional relational neural networks (CNNs), Deep Reinforcement learning and capsule networks.

The third circle is labelled neural networks, and expands out past generative AI and deep learning circles. It contains perceptrons, convolutional neural networks (CNNs), Multi-layer perceptron (MLP), backpropogation, activation functions, recurrent neural networks (RNN), self-orgnaizing maps (SOMs), dropout, Generative adversarial networks (GAN), and long short-term memory (LSTM).

The fourth circle is labelled Machine Learning. It contains feature engineering, ensemble learning, support vector machines, decision trees, dimensionality reduction, unsupervised learning, semi-supervised learning, reinforcement learning, classification, regression and clustering.

The fifth circle is labelled artificial intelligence, and it surrounds all of the previous circles. It includes cognitive computing, AI ethics, speech recognition, knowledge representation, planning and scheduling, natural language processing, computer vision, expert systems, robotics, automated reasoning and fuzzy logic.

GenAI tools help students generate ideas and improve writing by giving feedback.

You can also get help with writing from the Writing Centre and the Language Help Centre.

Generative AI tools like DALL-E and Adobe Firefly let users create digital artworks from descriptions. These tools may provide quick custom images to enhance projects.

Try using the Library's Research Overview tool within Page1+ for a reliable GenAI experience. This tool uses real, credible sources and avoids the common pitfalls of some GenAI tools such as fabricated data, ideas, and citations.

Watch the video Choosing Better Generative AI Tools (3:36 min) on YouTube for an overview of how to use this tool.

You can also get Research Help from the Library team.

GenAI tools can help to analyze large datasets in fields like data science and economics.

Generative AI (GenAI) can create tools to support students with disabilities including speech to text applications, real time captioning, and personalized learning aids.

Check in with your accessibility advisor or adaptive technologist for more details.

GenAI can sometimes produce information that is entirely fabricated or incorrect, known as "hallucination," which can lead to misinformation. For example, when asked about a historical event, GenAI might provide a detailed but entirely incorrect narrative, such as claiming a fictitious battle occurred between historical figures who never met.

GenAI’s quality depends on the training data; biases in data can lead to flawed information. For example, if trained on biased hiring data, GenAI might suggest predominantly male candidates for a tech job, reflecting existing gender biases in the industry.

Anything uploaded into a Generative AI system could potentially be made public by the software.  This means you should not upload proprietary, confidential or personal information.

Over-reliance on GenAI can hinder personal, academic and work-related skill growth. For example, a college student uses GenAI to complete their calculus homework, copying answers without understanding the steps. During exams, they struggle because they haven't developed the necessary problem-solving skills, leading to poor performance and gaps in their education.

GenAI may generate technically correct but impractical responses due to its lack of common-sense reasoning. For Example: When asked how to cook a meal in 20 minutes, GenAI will not know what level of cooking skill the user has, what type of equipment they have available to them, or what type of ingredients they can afford.

It struggles with context, missing nuances and intended tones in conversations. For example: When asked how much they should tip at a restaurant, GenAI may not know the quality of meal they had, the quality of service, or the local custom for tipping.

It simulates emotions but doesn’t truly understand them. For example: When asked for advice on what questions to ask an employer in an interview, GenAI will not be able to predict how an employer will behave or the responses they will give during the interview that may make those questions inappropriate.