Research reduces energy consumption through artificial intelligence, a promising change for e-commerce

Researchers at the University of California, Santa Cruz have developed a method to significantly reduce the energy costs associated with running large language models.

This is a development that may significantly impact the use of artificial intelligence (AI) in eCommerce. By reducing energy consumption, their approach could make advanced AI capabilities more accessible and affordable for businesses of all sizes.

“We achieved the same performance at a much lower cost by fundamentally changing the way neural networks work,” Jason Eshraghian, assistant professor of electrical and computer engineering at the Baskin School of Engineering at the University of California, Santa Cruz and lead author of the study: wrote in a Thursday (June 20) ) press release. “Then we took it a step further and built custom hardware.”

The cost of artificial intelligence in eCommerce

Currently, using advanced AI models like ChatGPT comes at a high price. Recent estimates suggest that OpenAI’s energy costs alone cost almost $700,000 per day. These costs translate into price and can create a significant barrier for smaller companies looking to leverage AI in their e-commerce operations.

Research by a team from the University of California, Santa Cruz aims to address the high energy costs associated with running advanced artificial intelligence models. By eliminating matrix multiplication, the most computationally expensive element of running large language models, they managed to increase the energy efficiency of the model.

“Neural networks are, in a sense, glorified matrix multiplication machines,” Eshraghian said. “The bigger the matrix, the more things your neural network can learn.”

The researchers say their approach is strikingly effective.

“We were able to power a billion-parameter-scale language model with just 13 watts of power, which is about the same amount of energy it takes to power a light bulb and more than 50 times more efficient than typical hardware,” Eshraghian said.

This level of performance could enable eCommerce platforms to offer advanced AI-powered features such as personalized recommendations, chatbots, and dynamic pricing at a fraction of the current cost.

Implications for e-commerce on mobile devices

The team’s innovation also has important implications for mobile e-commerce. Rui-Jie Zhu, first author of the paper and a graduate student in Eshraghian’s group, noted in a press release: “We have replaced expensive surgery with cheaper surgery.”

The reduction in computational complexity achieved by a team at the University of California, Santa Cruz could potentially enable full-scale artificial intelligence models to run on smartphones. This advancement comes at a time when mobile shopping is growing rapidly.

If implemented, this technology could significantly improve the mobile shopping experience and app-based e-commerce by enabling more sophisticated AI-based features, such as personalized recommendations and advanced search features, directly on users’ devices.

Taking advantage of advances in software, the team expanded their research by collaborating with other departments at the University of California, Santa Cruz to develop custom hardware. This specialized equipment has been designed to maximize the efficiency gains resulting from the new approach.

“These numbers are already really solid, but it’s very easy to improve them significantly,” Eshraghian said. “If we can do that with 13 watts, imagine what we could do with the computing power of an entire data center. We have all these resources, but let’s use them efficiently.”

For e-commerce giants with sprawling data centers, this could mean significant savings and improved AI capabilities. For smaller companies, this could level the playing field, allowing them to compete with more sophisticated AI-based strategies.

As PYMNTS previously reported, Big Tech companies like Microsoft and Google are struggling to profitably monetize their generative AI products due to the high costs of production, development and training.

As the eCommerce industry evolves, these types of innovations could change the way companies interact with customers, manage inventory and make strategic decisions. While the technology is still in its early stages, its potential to democratize advanced AI capabilities in the eCommerce sector is enormous.

The researchers have open-sourced their model, potentially accelerating adoption and further innovation in the field. As Eshraghian puts it, “we have fundamentally changed the way neural networks work.” The eCommerce world will be watching closely to see how this change translates into real-world applications and competitive advantage in the digital marketplace.