World's First Carbon-14 Diamond Battery, DeepMind's Genie 2, and Amazon's AI Mega-Computer
09:19 Share

Welcome to Discover Daily by Perplexity, an AI-generated show on tech, science, and culture. I'm Isaac, filling in for Alex today. And I'm Sienna. Imagine a battery that could power your great-great-great-grandchildren's devices. Well, researchers have just created one that could last for thousands of years. It's pretty fascinating. Researchers from the University of Bristol and the UK Atomic Energy Authority have been

have developed a battery that harnesses the radioactive decay of carbon-14. This isotope is encased in a diamond structure, which not only contains the radioactivity but also converts the energy from decay into usable electricity. That's incredible. What kind of applications are we looking at for this technology? It's vast,

Think about medical implants like pacemakers that would never need replacing, or spacecraft on long-duration missions that could be powered for decades without worrying about energy supply. The durability of these batteries also makes them ideal for use in extreme environments where conventional battery replacement is nearly impossible. But is it safe? This is radioactive material. That's a great question, and it's one of the key features of this technology.

The diamond structure effectively contains the radioactive material, ensuring safe operation. In fact, the researchers claim that the radiation levels are lower than the background radiation we're exposed to in everyday life. Plus, the durability of the diamond makes it incredibly resistant to damage. So what are the challenges in making this a reality on a larger scale? Well, there are a few significant hurdles.

First, the manufacturing process is quite complex. It requires specialized equipment, including a plasma deposition rig, and demands precise control over the diamond growth process. Then there's the regulatory aspect.

Handling radioactive materials comes with strict requirements, which add significant costs and complexity to the production process. And what about the economic factors? I imagine this isn't cheap to produce. Yeah, the production costs are influenced by the need for specialized safety infrastructure and complex manufacturing equipment. Currently, these batteries are significantly more expensive than traditional battery technologies. We're talking about $20,000 per microwatt,

which is a stark contrast to the $139 per kilowatt hour for lithium-ion batteries in 2023. It's likely that as the technology matures and production scales up, we'll see some reduction in costs.

But it's important to remember that these batteries are designed for very specific applications, where their long lifespan and durability justify the high initial cost. It's not about replacing your AA batteries, but rather powering critical devices in challenging environments for extremely long periods. So what should we be watching for as this technology develops?

First, we should watch for advancements in the manufacturing process that could bring down costs and increase efficiency. Companies like Arkenlight are already working on establishing production facilities near decommissioned nuclear power plants to facilitate easier access to carbon-14 isotopes. It's a clever approach to sourcing the necessary materials. Another thing to watch for is the first real-world applications of these batteries, particularly in medical devices and space exploration.

These fields could really benefit from the long-lasting, maintenance-free power these batteries provide. Speaking of cutting-edge technology, I heard there's some exciting news in the world of AI. Can you tell us about that? DeepMind, which is owned by Google, has just unveiled a new AI system called Genie 2. This is a foundation world model that can transform various inputs, like text prompts or sketches, into interactive 3D environments with realistic physics and spatial coherence.

Genie 2 uses a sophisticated AI model to interpret the input and generate a corresponding 3D environment. What's really impressive is its ability to create these worlds with realistic physics simulations and spatial coherence. This means you can actually interact with objects and navigate through the space in a way that feels natural and consistent.

The potential applications are quite diverse. In the gaming industry, it could revolutionize rapid prototyping, allowing developers to quickly create and test new environments. It could also enhance AI training by providing more realistic and varied virtual environments. Beyond that, there are potential applications in virtual reality, education, and even robotics. While Genie 2 is a significant advancement, it does have its constraints. The main limitation is the duration of interactivity,

Users can only explore the generated worlds for up to one minute. This obviously limits its utility for extended applications or gameplay scenarios. Another limitation is that while the fidelity of these environments is impressive, it may not yet match the level of detail and polish achieved by manually designed 3D worlds, especially in highly intricate or specialized settings. So what should we be watching for as this technology develops? First, watch how DeepMind addresses these limitations.

particularly in extending the duration of interactivity and improving the fidelity of the generated environments. It'll also be interesting to see how game developers and other industries start incorporating this technology into their workflows. As the technology matures, we might see more sophisticated applications in fields like urban planning, architectural visualization,

or even in creating virtual training environments for various professions. The ability to quickly generate and modify complex 3D environments could be a game changer in these areas. That's really exciting. There's also some big news in the world of AI computing power. Amazon Web Services and Anthropic are developing a groundbreaking AI supercomputer known as the UltraCluster. It's designed to be one of the most powerful AI training systems in the world.

utilizing hundreds of thousands of Amazon's proprietary Tranium 2 chips. That sounds massive. What makes this project so significant? Amazon claims that the UltraCluster will be five times larger and more powerful than the system Anthropic used to train its current AI models. This represents a substantial increase in computing power.

which could lead to significant advancements in AI capabilities. The UltraCluster will be utilizing AWS's latest Tranium 2 chips, which are specifically optimized for AI model training.

It's designed to handle advanced AI workloads, particularly for training large language models. One interesting aspect is its location. It'll be based in the United States, which positions it as a strategic asset for domestic AI development. That's interesting. How does this fit into the broader AI landscape? Well, this partnership represents a strategic move to challenge existing power dynamics in the AI industry. Amazon is investing a total of $8 billion in Anthropic,

and this collaboration positions both companies as major contenders in what some are calling the AI arms race. The UltraCluster offers several key advantages, including potentially reducing AI model training costs by up to 40% compared to existing solutions. This initiative directly challenges NVIDIA's dominance in the AI chip market.

NVIDIA currently holds about 95% market share in AI chips, so Amazon's move to develop its own AI chips and infrastructure is a significant disruption. It's worth noting that this isn't happening in isolation. We're seeing intense competition in AI infrastructure development across the tech industry. For example, Google's DeepMind is investing over $100 billion in AI development, and Elon Musk's XAI is expanding its Colossus supercomputer.

That's it for today. Thanks for tuning in and don't forget to subscribe on your favorite platform. For more info on anything we covered today, check out the links in our episode description. And don't forget, you can now access Perplexity's AI-powered knowledge base on the go with the mobile app, available for both Android and iOS. We also just released the Perplexity desktop app for macOS.

In other Perplexity news, we're excited to announce that Perplexity now offers a comprehensive one-stop shopping solution, where you can both research and purchase products. The platform now features Buy with Pro, a first-of-its-kind AI commerce experience offering one-click checkout and free shipping for Pro users in the US.

There's also an innovative snap to shop feature that lets you find products by simply taking a photo and an AI powered discovery system that provides unbiased product recommendations with clear visual product cards. The platform integrates with Shopify to access up to date product information from businesses across the US, making online shopping easier and more efficient than ever. We'll be back with more stories that matter. Until then, stay curious.