Public Seminar – Machine-Learning to Modern “AI” and Beyond
Machine-Learning to Modern “AI” and Beyond
(Connecting the Dots: How the Technology Works)
- Availability: Open to all Pawsey Staff and researchers
- Date: Tuesday (morning), 3rd of June, 2025
- Duration: 9:30AM → 12PM
- Location: Exhibition Space
- Delivery: online
Computing environments have forever changed with the advent of transformer as well as how humans interact with these AI models. This 90 minute talk will connect the dots from the first XOR models that demonstrated the computational universality of machine learning to the technology components that make up the transformer architecture. Key questions in how humans interact and realize value from these models will be examined along with the ever present need for verification. Recent trends highlight a divergence in approaches from large, expensive models that have to reside in the cloud to models that can run locally – even for thinking models. Innovators can exploit such capabilities to use AI as a collaborator, changing the mindset that they are a software tool, who can utilize domain-specific knowledge, design 3D parts, and train robots in exceeding fast physics-based simulation environments – even on a PC! Future trends such as quantum and neuromorphic computing will also be briefly touched upon.