The Interface

  1. 04/07/2025

    What’s the difference between reasoning and traditional AI models? Why is inferencing becoming cheaper? What’s next in AI? (Part 2)

    Taking the cue from the previous episode on the history of AI, all the way to ChatGPT, this episode looks into the concept of multi-modal AI. We explore how this technology integrates text, images, and audio to mimic human brain processing. We discuss fusion mechanisms that combine these modalities, allowing AI models to comprehend and respond to complex inputs. These mechanisms are crucial for practical applications, such as extracting information from PDFs or answering questions about images. Subsequently, we transition to reasoning models that can be prompted to provide sequential reasoning. Reasoning models, like DeepSeek’s r1, are designed to automatically reason through problems and manage the reasoning effort based on complexity. This approach distinguishes itself from prompting techniques such as “let’s think step by step” or “chain of thought,” which aim to enhance accuracy through structured reasoning. Group Relative Policy Optimization (GRPO) emerges as a reinforcement learning method employed to train models like DeepSeek R1. GRPO incentivizes model improvement through rewards, such as correct answers in mathematical problems. This approach facilitates self-supervised training without human intervention, enabling the emergence of extended thinking chains and enhanced responses. In the concluding segment of the discussion, we address the reduction in training and inference costs, even as companies invest substantial resources in GPUs for training large models efficiently. Algorithmic advancements and hardware improvements facilitate the training of smaller models, thereby increasing AI’s accessibility to enterprises and startups. Agentic AI, model context protocols, and smaller language models represent emerging trends that will shape the future of AI. These advancements will render AI more practical and efficient for real-world applications. Produced by Sharmada Venkatasubramanian

  2. 07/27/2024

    Ep 7 Is India blazing a trail on green data centres? with Vivek Dahiya

    Artificial intelligence has sparked an insatiable appetite for computing power and data storage, driving rapid expansion of data centres - the physical backbone of the digital world. At the forefront of this expansion are hyperscalers - data centres of tech giants like Google, Amazon, and Microsoft. Their massive, scalable data centre infrastructures power cloud services and AI applications that are becoming increasingly integral to business operations and daily life. As AI technologies continue to evolve and proliferate, the demand for data centre capacity is skyrocketing, leading to a global boom in data centre construction and investment. These modern data fortresses are growing not just in number, but in scale, with some facilities now consuming as much electricity as small cities. This growth, while enabling remarkable technological progress, also brings significant challenges. From the pressing need for sustainable energy solutions to the constant drive for greater efficiency and innovation, the intersection of AI, hyperscalers, and data centres represents a critical frontier in technology infrastructure - one that will play a pivotal role in shaping our digital future. India's data centre capacity is set to touch 2,000 MW by 2026. And the country is making plans to transition into a green economy. Joining me to explore the impact of data centres in India and in the Asia Pacific region is Vivek Dahiya, Managing Director & Head Data Centre Advisory Team, Asia Pacific, at Cushman & Wakefield. In this discussion, John and Vivek discuss:  How the emergence of AI changed the market for data centres globally? How different are developed and emerging markets in their approach towards data centre infrastructure? What is the impact of data centre on climate change?  What policies around green energy can make this industry green?   This episode is powered by AMD EPYC Processors. Host: John Xavier, Technology Editor, The Hindu Produced by Sharmada Venkatasubramanian