Princeton University has introduced an exciting new initiative to push the boundaries of artificial intelligence (AI) for academic and research purposes. The Princeton Language and Intelligence Initiative (PLI) has set its sights on deepening our understanding of large AI models, promoting their application across diverse academic disciplines, exploring AI’s societal and ethical implications, and devising strategies to mitigate any potential negative consequences.
At the core of PLI’s mission are Large Language Models (LLMs), sophisticated AI systems trained on extensive textual data to perform tasks like natural language processing, translation, and text prediction. These models have gained significant attention in recent years for their role in powering AI tools, including OpenAI’s ChatGPT.
The journey from conceptualizing PLI to its official launch took approximately a year. Notably, the initiative gained momentum thanks to the rising popularity of LLM-based chatbots, such as ChatGPT. Sanjeev Arora, the Charles C. Fitzmorris Professor of Computer Science, is leading this effort and is renowned for his expertise in theoretical computer science and machine learning.
PLI aspires to build a robust local computing infrastructure with equipment and personnel. This infrastructure will provide essential support for Princeton researchers, enabling them to harness the power of LLMs for various academic applications. According to Arora, this local computing capability could prove more cost-effective and legally viable for certain tasks, especially those involving proprietary data.
Empowering research across disciplines
PLI is not confined to computer science but seeks to engage students, researchers, and faculty members from various academic disciplines. This collaborative approach aims to unlock AI’s potential to address complex challenges across various fields.
In pursuit of its research-centered objectives, PLI may develop its LLMs customized to meet the specific requirements of Princeton researchers. These tailored models could open new avenues for AI-driven research and innovation across multiple domains.
Inaugural event: AI@Princeton
PLI’s inaugural AI@Princeton event took place at McCosh Hall 50 on September 26th. This event provided a platform for experts in computer science, psychology, classics, and other disciplines to delve into the workings of LLMs, their potential applications, and the associated societal risks and challenges.
During his presentation at AI@Princeton, Sanjeev Arora announced a substantial computing budget of $10 million for PLI. This financial commitment underscores Princeton University’s dedication to advancing AI research within its academic community.
PLI’s mission closely aligns with other Princeton groups, such as the Center for Digital Humanities (CDH), the Center for Information Technology Policy (CITP), and the Center for Statistics and Machine Learning (CSML). These collaborations reflect the interconnected nature of AI research and its potential to impact various academic domains.
University Provost Jennifer Rexford emphasized the need for PLI to adapt to the rapidly evolving AI landscape and establish a solid foundation before seeking external funding. Rexford noted that once the initiative gains momentum, it can explore support from federal funding sources, foundations, and other external partners.
High demand for advanced GPUs
PLI’s quest to acquire state-of-the-art graphics processing units (GPUs) underscores the surging demand for high-performance semiconductor chips. The demand for these GPUs, particularly the Nvidia H100 GPUs, has already exceeded supply, with orders placed before their production commences.
Meredith Martin, an associate professor of English and the Faculty Director of the Center for Digital Humanities (CDH), expressed optimism about PLI’s potential to accelerate research and innovation. The initiative’s commitment to attracting diverse faculty and fostering novel approaches to AI research promises to be a valuable asset for Princeton University.
Princeton’s PLI initiative represents a significant step towards harnessing the potential of AI for academic and research purposes. With a strong focus on LLMs and a robust local computing infrastructure, PLI is poised to drive innovation and collaboration across many academic disciplines. As it navigates the ever-evolving landscape of AI, PLI aims to position itself as a catalyst for cutting-edge research and ethical AI development within the academic community.
