#045 - India to Get World’s Fastest Computers AI Infrastructure from NVIDIA, The Hidden Cost of AI: Water Usage, Meta’s Ambitious Plan to Build a ChatGPT Competitor.
Fresh & Hot curated AI happenings in one snack. Never miss a byte 🍔
This snack byte will take approx 5 minutes to consume.
AI BYTE # 1 📢 : AI’s Thirst for Power: Why We Need to Regulate Water Consumption in Data Centers?
⭐ As AI systems like ChatGPT become more advanced and ubiquitous, they require more computing power and energy, which in turn means more water consumption for cooling the servers that run them.
According to a recent report by the Associated Press, AI systems like ChatGPT cause data centers to consume about 500 milliliters of water each time a user poses 5 to 50 prompts or questions.
That’s equivalent to half a liter of water for every few sentences generated by the chatbot. Considering that ChatGPT has been the fastest-growing tech site visited on earth since the creation of the internet, that’s a lot of water being used.
This water consumption is not only a waste of a precious resource, but also a potential threat to the environment and human health. The report cites the example of Iowa, where ChatGPT’s data centers are located near corn fields and use water from the Raccoon and Des Moines rivers.
The increased demand for water could affect the availability and quality of water for agriculture and drinking, as well as the ecosystems and wildlife that depend on it.
The problem is not limited to ChatGPT or Iowa. Microsoft and Google, the biggest players in the AI field, have also reported significant increases in their water consumption in recent years, driven by AI-related computing demands.
According to official reports, Microsoft’s water consumption spiked 34 percent from 2021 to 2022, while Google’s grew by 20 percent. That adds up to an additional 1.7 billion gallons of water between the two companies.
While tech companies claim to be taking steps to reduce their environmental impacts, such as using renewable energy, improving efficiency, and adopting ethical standards, experts warn that these measures may not be enough to keep up with the rapid growth of AI.
They urge more transparency, accountability, and regulation of AI development, as well as public awareness and education about its resource usage and potential consequences.
AI is undoubtedly a game-changer for many industries and applications, but it also poses serious challenges for sustainability and social justice.
We need to balance the benefits and risks of AI and ensure that it is developed and used in a responsible and ethical way. Otherwise, we may end up paying a high price for AI’s thirst for power.
AI BYTE # 2 📢 : Meta’s Ambitious Plan to Build a ChatGPT Competitor
⭐ Meta is making strides in the AI sector with plans to develop an AI system that rivals the capabilities of OpenAI’s ChatGPT.
This ambitious venture forms part of CEO Mark Zuckerberg’s strategy to establish Meta as a dominant player in the burgeoning AI industry.
The forthcoming AI model, which is anticipated to be more robust than Meta’s recent release, Llama 2, is being designed to facilitate the development of sophisticated text and analysis tools by other companies.
It’s reportedly being spearheaded by a special team initiated by Zuckerberg early this year, focusing on the creation of Generative AI instruments capable of crafting human-like expressions. The project is still in its infancy, with training scheduled to commence in early 2024.
Meta plans to funnel substantial resources to the project, including the construction of data centers and the acquisition of H100s, Nvidia’s most sophisticated chips designated for AI training.
Despite collaborating with Microsoft for Llama 2, Meta intends to utilize its own infrastructure for this venture, veering away from Microsoft’s Azure platform.
Zuckerberg is also still signaling that he intends the forthcoming AI offering to be open-sourced, making it freely accessible for companies to develop AI-driven tools.
This open-source ethos, however, brings with it a slew of potential pitfalls, including augmented risks associated with copyright infringements and the potential misuse of the technology for disinformation campaigns.
Despite these uncertainties, the industry recognizes the inherent benefits of open-source models, lauded for their cost-effectiveness and adaptability, as Zuckerberg avidly champions a collaborative approach to AI development in the rapidly evolving technological landscape.
What do you think of Meta’s new AI project?
Do you think it can compete with OpenAI’s ChatGPT? Let me know your thoughts in the comments below.
AI BYTE # 3 📢 - India to Get World’s Fastest Computers and AI Infrastructure from NVIDIA
⭐ India is poised to become one of the world leaders in AI, thanks to the partnership between NVIDIA, Reliance and Tata.
NVIDIA CEO Jensen Huang has revealed how his company is bringing the fastest computers in the world and the most advanced AI infrastructure to India.
According to Huang, India has everything it needs to build and use AI: data, talent, and now infrastructure.
He has said that by the end of next year, India will have access to AI supercomputers that are 50 to 100 times faster than anything the world has ever seen, lowering the cost and time of training foundational models.
He has said that Reliance and Tata will have access to DGX Cloud, a service that allows any business to use its own AI supercomputer through a regular web browser. This will enable them to create customer-centric AI applications and services for various domains, such as language, legal, HR, sales, etc.
Huang has also praised India’s diversity of languages and dialects, and urged the country to not export its data to Western companies. He said that India has the capability to make in-house large language models (LLMs) and foundational models using its own data and computer scientists.
He also mentioned that NVIDIA’s most advanced NVIDIA GH200 Grace Hopper Superchip and NVIDIA DGX Cloud would accelerate this process. He said that these technologies are not even in production yet, but India will be one of the first countries in the world to get them.
Huang is optimistic about India’s AI future. He said that if US companies like OpenAI required months to train GPT models, India would achieve similar results in a matter of weeks.
What do you think about this partnership and its impact on India’s AI ecosystem? Share your thoughts in the comments below.
