Grant: Discovery Launch Supplement
University of Toronto
$12,500.00
Developing and Enhancing User Systems using Adaptive Experimentation and Generative AI
FY 2024-25
Summary
The proposed research aims to improve the way we interact with software that integrates artificial intelligence algorithms and machine learning models to help us learn, work, and play. We want to make it easier for everyone, regardless of their technical or computer skills, to use emerging technologies such as ChatGPT. ChatGPT is a popular software application that uses a large language model (LLM) to provide answers to questions that users may ask. An LLM is a computer model of human language that was trained on massive quantities of example text (e.g., books, web articles, blog posts, Wikipedia, etc.). Software applications can send user questions or simple tasks to the model (e.g., "Please summarize this paragraph I have written into two sentences."), have it generate output, receive the output, and then provide the output to a user or use it to modify the application. When using applications such as ChatGPT today, users must ask the model questions using text-based prompts, which can be difficult to compose or correct if a response is unexpected. These challenges, in addition privacy concerns when asking questions and biases and misinformation within the models themselves, limit their accessibility and responsible use. Within this research, we want to make it easier for everyone to interact with such models and integrate them within the software applications you use every day.
We are thus focused on two problems. First, we need to better understand the current challenges that users face while creating prompts and integrating these models into tasks, workflows, and software in addition to the types of resources and support that people need while (i) initially learning to use such models and (ii) mastering use of them later. To develop this knowledge, we will create an interactive software platform that will allow us to rapidly test different types of support methods, feedback techniques, and the amount of support that users need while using current and future models.
Second, we need to develop methods to enable researchers to understand the differences between LLMs and smaller language models (SLMs) that have fewer privacy issues and were trained on smaller, focused datasets so they have less misinformation and bias (e.g., news articles specifically about chemistry rather than ever news article available online). These methods will enable researchers to compare the performance, adaptability, and responsiveness of these models and provide the data needed for researchers and software developers to create more responsible, user-focused software with them.
By accomplishing these two goals, we will uncover much needed knowledge about how software developers should design future software to best support our interactions with emerging technologies while also enabling these technologies to increase the usability and enjoyment we have with the mobile phones, computer software, and entertainment devices that we use every day.
Program
Discovery Launch Supplement
Recipient
Location
Ontario, Canada
Recipient Type
Academia
Type
Grant
Source
NSERC's Awards Data
As published
Application id
DGECR-2024-00485
Name nom
Liut, Michael
Department département
Mathematical and Computational Sciences
Organization id
31
Institution établissement
University of Toronto
Province en
Ontario
Province fr
Ontario
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2024
Competition year année de concours
2024
Award amount
12500
Program id
DGECR
Program name en
Discovery Launch Supplement
Program name fr
Supplément Tremplin vers la découverte
Group en
DISCOVERY RESEARCH
Group fr
RECHERCHE AXÉE SUR LA DÉCOUVERTE
Committee code
1507
Committee name en
Computer Science
Committee name fr
Sciences informatiques
Area of application code
802
Area of application group en
Information and communication services
Area of application group fr
Services d'information et de télécommunications
Area of application en
Information systems and technology
Area of application fr
Systèmes d'information et technologie de l'information
Research subject code
2700
Research subject group en
Information technology
Research subject group fr
Technologie de l'information
Research subject en
Information technology
Research subject fr
Technologie de l'information
Field of research list names en
Not available
Field of research list names fr
Non disponible
Socioeconomic objective list names en
Not available
Socioeconomic objective list names fr
Non disponible
Application title
Developing and Enhancing User Systems using Adaptive Experimentation and Generative AI
Keywords
adaptive interfaces; adaptive systems; generative ai; large language models; small language models; scaffolding learning; human-computer interaction; machine learning
Application summary
The proposed research aims to improve the way we interact with software that integrates artificial intelligence algorithms and machine learning models to help us learn, work, and play. We want to make it easier for everyone, regardless of their technical or computer skills, to use emerging technologies such as ChatGPT. ChatGPT is a popular software application that uses a large language model (LLM) to provide answers to questions that users may ask. An LLM is a computer model of human language that was trained on massive quantities of example text (e.g., books, web articles, blog posts, Wikipedia, etc.). Software applications can send user questions or simple tasks to the model (e.g., "Please summarize this paragraph I have written into two sentences."), have it generate output, receive the output, and then provide the output to a user or use it to modify the application. When using applications such as ChatGPT today, users must ask the model questions using text-based prompts, which can be difficult to compose or correct if a response is unexpected. These challenges, in addition privacy concerns when asking questions and biases and misinformation within the models themselves, limit their accessibility and responsible use. Within this research, we want to make it easier for everyone to interact with such models and integrate them within the software applications you use every day.
We are thus focused on two problems. First, we need to better understand the current challenges that users face while creating prompts and integrating these models into tasks, workflows, and software in addition to the types of resources and support that people need while (i) initially learning to use such models and (ii) mastering use of them later. To develop this knowledge, we will create an interactive software platform that will allow us to rapidly test different types of support methods, feedback techniques, and the amount of support that users need while using current and future models.
Second, we need to develop methods to enable researchers to understand the differences between LLMs and smaller language models (SLMs) that have fewer privacy issues and were trained on smaller, focused datasets so they have less misinformation and bias (e.g., news articles specifically about chemistry rather than ever news article available online). These methods will enable researchers to compare the performance, adaptability, and responsiveness of these models and provide the data needed for researchers and software developers to create more responsible, user-focused software with them.
By accomplishing these two goals, we will uncover much needed knowledge about how software developers should design future software to best support our interactions with emerging technologies while also enabling these technologies to increase the usability and enjoyment we have with the mobile phones, computer software, and entertainment devices that we use every day.