AI as a Foundational Tool: Dr. Rose Yu on Transforming Business and Science

Oceanside Perspective
Jun 16
4 min read

AUTHOR | Chihyu (Linda) Liu

Executive Summary

Dr. Rose's presentation illuminated the pervasive and increasing influence of Artificial Intelligence (AI) across critical sectors, emphasizing its shift from niche applications to a fundamental technology driving performance and discovery. Highlighting examples from early business process automation at USPS to modern supply chain optimization, the talk underscored AI's role in separating high-performing organizations. Furthermore, Dr. Rose showcased cutting-edge research applying AI, particularly deep learning, to solve complex, real-world problems in traffic forecasting, global climate modeling, epidemic response, economic cause discovery, and even enhancing the numerical capabilities of language models, establishing AI's essential role in the future of engineering, science, and commerce.

AI's Journey in Business: From Sorting Mail to Optimizing Supply Chains

Artificial Intelligence is not a new phenomenon in the business world. Dr. Rose pointed to early, practical applications, such as the US Postal Service leveraging AI for handwritten address recognition to categorize mail efficiently—a system used for over two decades. This historical context sets the stage for AI's current, more sophisticated role.

In contemporary supply chain management, AI/ML is proving to be a significant differentiator. Dr. Rose shared compelling data indicating that companies leveraging AI to automate and optimize key processes—demand forecasting, order management, supply planning, logistics, and S&OP—demonstrate markedly higher performance. Notably, high performers utilize AI in these areas at roughly double the rate of lower performers, clearly linking AI adoption to operational success.

Tackling Grand Challenges: AI in Science and Engineering

While AI excels at pattern recognition, predicting non-physical phenomena presents unique challenges. Dr. Rose detailed several ambitious projects her team is undertaking to push these boundaries:

Revolutionizing Traffic Forecasting: Traditional traffic management relied on rigid, non-data-driven models (like mixed-integer programs). The advent of real-time sensing technology, combined with deep learning, has transformed this field. Dr. Rose's work enables accurate traffic forecasts extending up to an hour, a significant leap from the previous 5-10 minute horizon. This advanced capability is now integral to widely used platforms like Google Maps.
Accelerating Global Climate Science: Understanding the complex impacts of emissions, EV adoption, and policy changes traditionally required massive physics-based simulations taking up to six months. Dr. Rose's team, collaborating with climate scientists and Google DeepMind, developed a "Fusion based" deep learning model. This AI emulator drastically speeds up the process, successfully modeling global climate dynamics at a six-hour resolution, stably, for 100 years—a first in the field—enabling faster insights for a rapidly changing world.
Enhancing Epidemic Response: During the recent pandemic, Dr. Rose collaborated with the CDC. Her team developed AI models capable of simulating the effects of interventions like vaccine allocation or lockdowns, reducing scenario generation time from a week to a single day. These tools also enabled month-long pandemic trajectory forecasting based on historical data, aiding public health decision-making.

Unlocking Deeper Insights and Expanding AI's Capabilities

Beyond prediction, Dr. Rose's research delves into understanding underlying causes and enhancing AI itself:

Discovering Hidden Economic Drivers: Identifying the root causes behind phenomena ("Causal Discovery") has traditionally been slow using methods like genetic algorithms. Dr. Rose introduced a novel deep learning-based Causal Discovery algorithm that offers significantly improved speed and scalability. This tool is already yielding interesting insights into the drivers of financial market shifts, stock price fluctuations, and potentially even large-scale climate patterns like the North Atlantic Oscillation.
Improving Language Model Numeracy: Acknowledging a key weakness in current Generative AI, Dr. Rose is tackling the challenge of LLMs' poor handling of numbers. Her team is developing a hybrid framework that injects physics knowledge and numerical simulations into these models. The goal is to create AI that comprehends both text and numbers effectively, paving the way for more robust applications in business problem-solving.

Insights and Takeaways

From a manufacturing perspective, I believe the future is full of promise, and there is reason for optimism. The integration of AI into supply chain operations—particularly in logistics and route optimization—has the potential to significantly enhance efficiency. In my previous role at a B2B food manufacturing company, we developed a cost management business intelligence (BI) system and continuously refined our delivery routes over a five-year period. As a result, four out of seven sales teams eventually became profitable.

Dr. Rose’s work in real-time sensing and deep learning enables traffic forecasting up to an hour in advance—an advancement that can greatly improve delivery route planning and overall network optimization. These predictive logistics capabilities enhance both responsiveness and operational efficiency.

For businesses, more accurate forecasting translates directly to reduced labor costs associated with traffic delays and increased delivery turnover. Improved operational efficiency, in turn, generates higher profits, which can be reinvested into new technologies—creating a self-reinforcing cycle of innovation and growth.

Conclusion: AI as a Fundamental Pillar

Dr. Rose's overview painted a clear picture: AI is no longer just an auxiliary tool but a fundamental component driving advancements across computing, engineering, business, and scientific research. From optimizing established processes to enabling breakthroughs in complex system modeling and causal understanding, AI provides the capabilities needed to address intricate challenges and unlock new levels of performance and insight. The continued development and application of these technologies promise to further reshape our world.

Author's Bio

Chihyu (Linda) Liu

Co-President at Operations and Supply Chain Club, UC San Diego

Full-time MBA candidate, UC San Diego, Rady School of Management

Linda is an MBA candidate at UCSD's Rady School of Management focusing on Supply Excellence and Innovation, currently serving as Co-President of the Operations and Supply Chain Club. She brings over five years of experience from Sunright Food Corporation, Taiwan's largest miscellaneous food manufacturer, holding roles in Product Management and as a Trading Specialist.

At Sunright, Linda demonstrated strong analytical and leadership skills, driving significant business growth by doubling the B2B client base and increasing key client revenue by 30% through data analysis and targeted digital marketing. She successfully optimized supply chain operations via strategic procurement and logistics management, achieving notable cost savings, and led cross-functional teams in launching multiple new products based on market analysis. Linda excels at leveraging data and collaboration to manage complex projects and achieve strategic objectives.