“We are at a once-in-a-generation moment,” said James McNiven, Arm’s Vice President of Product Management, when he took the stage as the first presenter with his “The Scalability of Intelligence: What the Future of AI Demands” forum at the afternoon session on May 21 at Taipei Nangang Exhibition Center Hall 2.
Despite declaring himself an “AI optimist,” he highlighted a risk that the scalability of AI would be compromised unless continuous innovation addressed the growing demand, in particular since the boom of GenAI two years ago.
“We are already on a collision course with reality,” McNiven warned, pointing out a ChatGPT interaction could use up to 15 times more energy than a traditional search query, while ChatGPT alone was estimated to consume as much power as a small US city every day.
Innovation in AI models could enable edge AI application for the sake of privacy, latency, cost and efficiency, democratization and availability, he said, “Let’s build an AI ecosystem that’s powerful, efficient and fair for everyone.”
Taking the stage second was Saurabh Kulkarni, Intel’s Vice President of AI Systems and GPU Products, with his “The Open Path to Enterprise AI: Flexible, Reliable, and Performant Solutions” forum.
The rise of agentic AI implied that it was moving beyond automating simple task and created complex pipelines that connected AI agents working in unison and incorporated decision making, he said.
Intel approached this challenge with a holistic approach that prioritized software and systems, he said.
Echoing with McNiven, he said Intel aimed to break down barriers and scale possibilities, making AI acceleration accessible to everyone.
He announced the public availability of the AI Assistant Builder on GitHub, a free tool to enable the creation of locally deployable AI agents on PCs.
“Creativity for all,” said Jeremy Willmott, Adobe’s Director of Digital Strategy Group, APAC, in his “AI powered Customer Experience Orchestration” forum, emphasizing that “creativity is the new productivity.”
Delivering great experiences at scale is not feasible without AI, he said, referring to reports that showed 89 percent of executives believe they would not reach business goals in the next three years without scaling AI.
Adobe’s AI platform unleashed creativity at an unprecedented scale, with over 22 billion images and content pieces having been created using Firefly models launched just 24 months ago, he said.
Adobe’s mission was to change the world through personalized experiences and empower everyone, from the biggest brands to individuals to tell their story, he said.
The last forum “New Millennium of AI-accelerated Engineering and Science” was presented by Ben Gu, Cadence’s Corporate Vice President of Multiphysics System Analysis.
He said the Millennium M2000 GPU acceleration solution, announced earlier this month and developed in close collaboration with Nvidia engineers to optimize algorithms in the latest Blackwell architecture, had achieved amazing results.
It achieved up to 80 times performance improvement on solvers and up to 20 times power reduction, he said, describing it as “tremendous and unbelievable.”
He quoted Nvidia’s CEO Jensen Huang as saying that “Millennium will transform discovery across disciplines.”
