Partnership Aims to Push Generative AI Research Forward to Help Enterprises Solve their Most Computationally Complex Problems with Quantum and Classical Algorithms.
“We’re thrilled to partner with an institute with such a longstanding reputation for innovation in the field,” said Yudong Cao, CTO and co-founder at Zapata AI. “Quantum algorithms running on classical computers can deliver tangible value for generative AI applications, particularly those that are time consuming and expensive to run. We’re looking forward to working with Mila researchers to refine these algorithms and help our enterprise customers run their generative AI models faster, cheaper, and at a wider scale — both for language-based applications and those involving numerical or time-series data.”
Together, Zapata AI and Professor Guillaume Rabusseau, Assistant Professor at Université de Montréal and Canada CIFAR AI Chair, alongside other members of the Mila community, will vastly broaden their resources and deep research capabilities that support the development of generative AI applications for industries such as life sciences, financial services, discrete manufacturing, and heavy machinery. With these combined efforts, the teams are aiming to discover new potential AI applications and mathematical formulations that could help enterprises solve complex business problems.
“This partnership with Zapata AI allows some of our researchers to combine emerging quantum and quantum-inspired techniques with state-of-the-art classical methods,” said Stéphane Létourneau, Executive Vice-President of Mila. “With our overlapping interests in core areas including health, environment and climate change, and AI ethics, our collaboration was a natural fit.”
In the coming weeks, Zapata AI and Mila will publish the first research paper born from their collaboration, which builds on Zapata AI’s previous research on quantum-inspired generative models. The research enables these models to learn from data on continuous variables, which include images, time-series data, and real-world sensor data. These variables are common in industrial problems and this research could be applied to generate new and improved solutions to these problems.
SOURCE: BusinessWire
