The race to harness the power of Generative AI (Gen AI) is on, and the aviation industry is no exception. American Airlines, a frontrunner in innovation, is exploring how Gen AI can revolutionize various aspects of its operations, from personalized passenger experiences to predictive maintenance. However, deploying Gen AI models on traditional CPUs requires a focus on computational efficiency and resource utilization. This is where Numenta's software optimization solutions offer a potential game-changer for American Airlines' Gen AI ambitions.
CPUs and the Growing Demands of Gen AI
Gen AI models, known for their ability to generate text, images, and code, are data-hungry and computationally intensive. While GPUs have their place on specific tasks, their high cost and power consumption often make them less suitable for large-scale deployments like those required by airlines. Conversely, CPUs are ubiquitous and more cost-effective and can provide an important option for Gen AI solutions.
Tailoring CPUs for Gen AI Efficiency
Numenta's software optimization technology addresses this challenge by analyzing and optimizing Gen AI models specifically for CPU architectures. Their techniques, such as quantization and sparsity, reduce the model's size and computational complexity without sacrificing accuracy. This translates to:
- Faster inference: Optimized models run significantly faster on CPUs, enabling real-time Gen AI applications at the scale American Airlines requires.
- Reduced resource consumption: Numenta’s optimized models demand less memory and 5-21x less power, leading to lower operational costs and potentially enabling deployment on edge devices like tablets or onboard computers.
- Improved accessibility: By democratizing Gen AI on CPUs, Numenta's solutions make it more accessible for airlines like American, regardless of their existing infrastructure investments.
“Numenta’s software optimization presents a unique opportunity for us,” says Tassio Carvalho Sr. Manager Operations Research and ML/AI at American Airlines. “Our strategy is to scale up our Gen AI PoCs quickly. The work with Intel and Numenta is key to achieving that goal”. By unlocking the potential of Gen AI on readily available hardware, American Airlines can revolutionize various aspects of their operations, creating a more personalized and efficient travel experience for their passengers.
Numenta's optimization will ensure these capabilities are generated efficiently on the airline's existing CPU infrastructure, delivering a seamless, private, and personalized experience without straining resources. Moreover, the cost-effectiveness of CPU-based Gen AI will allow American Airlines to scale Gen AI solutions across their business, i.e. entire passenger base, fostering brand loyalty and satisfaction.
“True to our Vision statement: "Innovate like a startup and Scale like an Enterprise." as part of our "Cloud Platform Innovation" and "Moonshot Thinking" with a partnership with Intel, we can bring in innovative solutions such as Numenta to enable our Application team such as Operations Research and ML/AI Ops to generate valuable business outcomes to American Airlines” noted Vijay Premkumar, Sr.Manager, Cloud and Platform Engineering.
Subutai Ahmad, Numenta CEO further explains: “While many companies are in the early stages of their Generative AI journey, American Airlines is well-positioned to extend their industry leadership simply by running AI applications on their existing infrastructure. The products we’re building at Numenta are designed to leverage the flexibility of CPUs while providing superior performance.”
As the aviation industry embraces the transformative power of Gen AI, Intel, and Numenta's solutions can serve as a powerful tailwind, propelling American Airlines to new heights of innovation and customer satisfaction.
Dr. Melvin Greer is an Intel Fellow and Chief Data Scientist, Americas, Intel Corporation. He is responsible for building Intel’s data science platform through graph analytics, machine learning, and cognitive computing. His systems and software engineering experience has resulted in patented inventions in Cloud Computing, Synthetic Biology, and IoT Bio-sensors for edge analytics. He is a principal investigator in advanced research studies, including Distributed Web 3.0, Artificial Intelligence, and Biological Physics. Dr. Greer serves on the Board of Directors, of the U.S. National Academy of Science, Engineering, and Medicine. Dr. Greer has been appointed and serves as Senior Advisor and Fellow at the FBI IT and Data Division. He is a Senior Advisor at the Goldman School of Public Policy, University of California, Berkeley, and Adjunct Faculty, at the Advanced Academic Program at Johns Hopkins University, where he teaches the Master of Science Course “Practical Applications of Artificial Intelligence”.