Canada Spends Logo

Explore our Budget 2026 analysis

Grant: Research Tools and Instruments

Dalhousie University

$16,662.00

Trustworthy Retrieval Augmented Generation for General Population Facing Services

FY 2024-25

Summary
The trustworthiness of Artificial Intelligence (AI) models has raised concerns due to their inadequate transparency and concealed predictive bias, posing challenges in deploying them in general population-facing services, such as healthcare, banking, and social media. Foundation models trained on Web-scale big data, despite their remarkable reasoning ability, are often inevitably laced with private or misleading information, and even anti-social content such as racial hatred, causing substantial research challenges to investigate. The objective of this RTI proposal is to support my research group with the hardware necessary to enhance the trustworthiness of AI-driven information services (in particular, Retrieval Augmented Generation) from multiple aspects, including prediction explanation, social bias mitigation, factual faithfulness enhancement, and further evaluation criteria exploration. The research directions involve large foundation models (with billions of parameters) and corresponding analysis modules (e.g., counterfactual testing modules or local prediction approximation modules) that cause additional computation overheads. Computational limitations we are facing usually come from insufficient GPU computing power and memory; loading a modestly sized foundation model into GPU can take 16GB memory minimum, beyond the capacity of our existing accessible resources inapplicable. I propose acquiring three GPU workstations equipped with Nvidia RTX 4090 GPUs. This equipment will be crucial to my research program's success, primarily funded by one NSERC Discovery grant, four Mitacs Accelerate, two NSERC IRAP grants, and one Nova Scotia Productivity and Innovation Voucher Program (PIVP). There is no financial overlap between this proposal and the student training grants listed above. However, there is substantial conceptual overlap over the research focus, and without the proposed equipment, our research will be severely delayed. The proposed equipment will provide state-of-the-art computing resources and hands-on training to all members of my research group. This will give them highly valuable skills in high-performance and parallel computing, alongside indirect skills, including software development, operations, engineering, and testing. The outcome of this research will be algorithms, techniques, tools, and best practices that enable continued trustworthiness improvements in the general population facing AI applications. This research also supports safety-critical AI applications that we and our industry partners value most, including healthcare, Banking, social media, and Marketing. Scientists, engineers, and content creators will be able to tackle more significant and complex decision-making challenges than ever before.
Program
Research Tools and Instruments
Location
Nova Scotia, Canada
Recipient Type
Academia
Type
Grant
Source
NSERC's Awards Data

As published

Application id
RTI-2025-00117
Name nom
Wu, Ga
Department département
Faculty of Computer Science
Organization id
79
Institution établissement
Dalhousie University
Province en
Nova Scotia
Province fr
Nouvelle-Écosse
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2024
Competition year année de concours
2025
Award amount
16662
Program id
RTI
Program name en
Research Tools and Instruments
Program name fr
Outils et instruments de recherche
Group en
DISCOVERY RESEARCH
Group fr
RECHERCHE AXÉE SUR LA DÉCOUVERTE
Committee code
1607
Committee name en
Computer, Mathematical and Statistical Sciences (RTI)
Committee name fr
Sciences informatiques, mathématiques et statistiques (OIR)
Area of application code
801
Area of application group en
Information and communication services
Area of application group fr
Services d'information et de télécommunications
Area of application en
Computer software
Area of application fr
Logiciel
Research subject code
2800
Research subject group en
Artificial intelligence
Research subject group fr
Intelligence artificielle
Research subject en
Artificial intelligence (computer vision, use2603)
Research subject fr
Intelligence artificielle (vision artificielle, utiliser 2603)
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
Trustworthy Retrieval Augmented Generation for General Population Facing Services
Keywords
machine learning; ai safety; retrieval augmented generation; trustworthy ai; deep learning
Application summary
The trustworthiness of Artificial Intelligence (AI) models has raised concerns due to their inadequate transparency and concealed predictive bias, posing challenges in deploying them in general population-facing services, such as healthcare, banking, and social media. Foundation models trained on Web-scale big data, despite their remarkable reasoning ability, are often inevitably laced with private or misleading information, and even anti-social content such as racial hatred, causing substantial research challenges to investigate. The objective of this RTI proposal is to support my research group with the hardware necessary to enhance the trustworthiness of AI-driven information services (in particular, Retrieval Augmented Generation) from multiple aspects, including prediction explanation, social bias mitigation, factual faithfulness enhancement, and further evaluation criteria exploration. The research directions involve large foundation models (with billions of parameters) and corresponding analysis modules (e.g., counterfactual testing modules or local prediction approximation modules) that cause additional computation overheads. Computational limitations we are facing usually come from insufficient GPU computing power and memory; loading a modestly sized foundation model into GPU can take 16GB memory minimum, beyond the capacity of our existing accessible resources inapplicable. I propose acquiring three GPU workstations equipped with Nvidia RTX 4090 GPUs. This equipment will be crucial to my research program's success, primarily funded by one NSERC Discovery grant, four Mitacs Accelerate, two NSERC IRAP grants, and one Nova Scotia Productivity and Innovation Voucher Program (PIVP). There is no financial overlap between this proposal and the student training grants listed above. However, there is substantial conceptual overlap over the research focus, and without the proposed equipment, our research will be severely delayed. The proposed equipment will provide state-of-the-art computing resources and hands-on training to all members of my research group. This will give them highly valuable skills in high-performance and parallel computing, alongside indirect skills, including software development, operations, engineering, and testing. The outcome of this research will be algorithms, techniques, tools, and best practices that enable continued trustworthiness improvements in the general population facing AI applications. This research also supports safety-critical AI applications that we and our industry partners value most, including healthcare, Banking, social media, and Marketing. Scientists, engineers, and content creators will be able to tackle more significant and complex decision-making challenges than ever before.