[SAIT AIRC AI Seminar] LLM & genAI - Technology, Business, and AI Markets

Abstract

The rapid advancement of deep learning (DL) technology has catalyzed the proliferation of AI applications across diverse domains such as computer vision (CV), natural language processing (NLP), recommendation systems (RecSys), and reinforcement learning (RL). These advancements have profoundly influenced various facets of daily life, fueling competition in self-driving car development, facilitating customer service through chatbots, generating creative content, and enhancing e-commerce experiences through personalized recommendations.

Despite these strides, the successful integration of machine learning (ML) techniques into industrial sectors has been limited, hindered by numerous challenges. Manufacturing environments, for instance, are characterized by data drift and stringent accuracy requirements, posing obstacles for conventional ML methods. Additionally, industrial applications often lack accurate labels or rely on subjective human judgments, further complicating the deployment of AI solutions.

This seminar introduces industrial AI as a burgeoning field with vast potential to deliver tangible business benefits and streamline engineering processes. Through case studies, we illustrate successful ML implementations in challenging industrial contexts, highlighting strategies for overcoming barriers. We delve into time-series ML algorithms and their pivotal role in manufacturing, emphasizing their relevance in handling the ubiquitous time-series data generated on production lines.

The second part of the seminar covers rapidly developing AI areas: LLM and genAI. The contemporary landscape of artificial intelligence (AI) has undergone a profound transformation, driven by the emergence and widespread application of large language models (LLMs), with OpenAI’s ChatGPT based on GPT-x models serving as a pivotal catalyst. The latter part of the seminar, tailored for engineering, managers, and decision-makers in virtually all industry sectors, delves into the dynamic intersection of LLMs and generative AI (genAI), engaging both major industry players and agile startups in a comprehensive exploration.

As intriguing as LLM and genAI are, the AI landscape is also witnessing dynamic shifts in trends and business impacts, accompanied by a surge in startup investments. The nature, development, and commercialization of LLM foundation models, as well as the niche opportunities for many AI startup companies, are thoroughly analyzed, adding a layer of excitement to the evolving AI ecosystem.

Beyond the captivating advancements of LLM and genAI, the AI landscape is undergoing dynamic transformations in both trends and business implications, accompanied by a notable surge in investments within the startup ecosystem. Delving deeper, we scrutinize the intricate nature, development, and commercialization of LLM foundation models, alongside identifying niche opportunities for burgeoning AI startups, infusing the evolving AI ecosystem with an additional layer of anticipation and opportunity.

Moreover, our discussion extends to the burgeoning changes within the hardware industry, particularly the intensifying competitive landscape among manufacturing giants such as TSMC, Nvidia, Intel, Global Foundry, Samsung, and SK Hynix. This competition underscores the pivotal role of hardware innovation in shaping the future trajectory of AI, with profound implications for various sectors. Additionally, we explore the concerted efforts by the United States to revitalize semiconductor manufacturing capabilities, reflecting a strategic imperative to bolster national technological prowess and security in an increasingly competitive global landscape.

Claude’s abstract

This presentation explores the transformative potential of Industrial AI in manufacturing, with a particular focus on semiconductor fabrication environments. The research demonstrates how fast AI adoption creates significant competitive advantages, with front-runner organizations achieving +122% cash flow improvements by 2030 compared to -23% for laggards. Semiconductor manufacturing serves as an ideal starting point for Industrial AI deployment due to its advanced digitalization infrastructure, generating massive data volumes (~1TB/day from equipment sensors, ~10TB/day from metrology images) and sophisticated processes that enable scalability to other industrial sectors.

The technical implementation encompasses two primary AI applications: computer vision for metrology and inspection, and time-series machine learning for process optimization. Computer vision techniques address challenges in scanning electron microscope (SEM) image analysis, including shot noise reduction, automatic critical dimension measurement with <0.1nm precision, and unsupervised anomaly detection. Time-series ML applications focus on virtual metrology, yield prediction, and root cause analysis, utilizing semi-supervised learning approaches with Bayesian inference to handle credibility intervals and address the inherent challenges of covariate shift, concept drift, and fat data scenarios prevalent in manufacturing environments.

The research presents a successful Virtual Metrology (VM) system that predicts unmeasured material properties using equipment sensor signals, achieving measurement precision comparable to physical metrology equipment while providing credibility intervals for prediction reliability. This breakthrough enables manufacturers to virtually measure all processed materials—equivalent to investing in 100x measurement equipment—and optimally reallocate limited measurement resources. The business impact includes potential tens of millions of dollars in savings through 1% yield increases and significant equipment cost reductions, demonstrating the substantial economic value of properly implemented Industrial AI solutions in manufacturing operations.