Google Developer Experts’ Post

Google Cloud Next'24 in Apr 9-11 @Las Vegas. Here are our real stories that how GenAI has helped drive tangible business outcomes. In Daangn case, we will discover how Karrot (= Daangn), a hyperlocal super app, is revolutionizing local marketplace, community engagement using GenAI in Google Cloud. This session highlights our journey in seamlessly integrating Gemini to construct a GenAI based modern knowledge extraction system, and how we use this wealth of data to improve our recommendation services. With NEXON’s AI project,   we will show customized AI models to detect harmful images uploaded into the NEXON ecosystem by its users. It benefits: 1) Higher accuracy of harmful image detection leads to better user experience , 2) Reduced operational burden (less number of user requests/flags to take down "harmful" or "disgusting" images).

OpenAI is sprinting ahead to establish a network effects moat around their models. Their focus on function calling and plugin architectures could potentially transform OpenAI into the new 'AI Cloud.' 🌐☁️ However, the race to build network effects is still in its early stages, and it's too soon to crown a winner. 🏁👑 There's a whole lineup of contenders extending this concept to create agents like LlamaIndex, Langchain, AutoGPT, BabyAGI, and more. Their goal? To automate aspects of your app, infrastructure, or even your daily life. 📱🏗️🔄 #OpenAI #AICloud #Automation

Announcing a preview of Cloud TPU v5p—the most powerful, scalable, and flexible TPU to date—which is now powering AI Hypercomputer, an accelerated supercomputer architecture that features an integrated system of performance-optimized hardware, open software, leading ML frameworks, and flexible consumption models TPUs are #AI accelerators that have long been at the ❤️ of Google's AI-powered products that serve billions of users like YouTube, Gmail, Google Maps, Google Play, and Android. In fact, Gemini, our most capable AI model, was also trained at scale on Cloud TPUs. Learn more here → https://goo.gle/47JPeV8

Build Your LLM-based app with Google Cloud Vertex AI and TruLens at the ⭐️TruEra Challenge⭐️. Make the most of the Zilliz vector database for your LLM-based app to receive direct support and increase your app's odds of success! [ Online Hackathon, December 1st - 8th ] 🛠️ Zilliz offers a cutting-edge Vector Database tailored for Enterprise-grade AI and LLM applications. This advanced technology serves as a robust foundation, empowering developers to optimize their AI projects with efficiency and scalability. 💎 Team members will receive $30 in Google Cloud credits, with an additional $100 in Google Cloud credits awarded after they submit their solution 🏆 $9,000 Cash Prize Pool and $14,000 in Google Cloud credits for the winners Don't wait any longer, sign up to save your spot: https://bit.ly/47IunBf ⬅️ lablab.ai, Google, Google Cloud, Google Cloud Partners, TruEra, Zilliz, LlamaIndex, Intel Corporation, Intel AI

Web LLM and Web Stable Diffusion blew my mind! They run entirely in your browser without an internet connection. How? By using WebGPU in Chrome! They're open-source and hint at a future with locally run AI models. Soon, you won't need cloud APIs for tasks like summarization or translation – it'll all happen right on your device.

If you're like me, you wished you could have gone to #Google #Next but missed it. To say the conference focused on #AI would be an understatement. After seeing the announcements and watching this video, it becomes all too clear we are in the grand AI battle royale between the major cloud providers and numerous smaller players. Check out this video for a quick overview of the Next 2024 Keynote, and let the AI games begin! https://lnkd.in/gw9UWBAC

Huggingface & Github are exploding with new AI tools. I've tried exploring video creation these past couple of days (of easter) and these tools require no cloud. Everything runs perfectly fine in the home lab. I guess most people have seen the Sora previews, but almost nobody has access yet. The video clips I've been able to generate are quite short, and the characters are not consistent. I've tried taking a robot around the world to explore the possibilities. The tools used here are VideoCrafter2 (by TenCent) and MAGNeT (by Meta).

My new fave AI tool, Pieces has a VS Code extension, which has joined my list of must-haves. The Copilot mode supports multiple Cloud and on-device LLM including the new GPT4o. You can analyse your code and save it as a snippet in the Pieces app. https://lnkd.in/eUvCkF_t

While many of the #AI announcements at #Google #Next were incremental, one of the biggest takeaways is the significant #investments Google has made to the AI platform for making it very easy to build AI solutions even without code.

With my work in Marrow and Rebar I'm constantly reading about developments in AI technology to try to keep up with the latest. Frankly things are happening so fast that it's really difficult. So I also listen to smart people who are also paying attention like Chamath. I think his take on commoditization of models is pretty accurate. In our work we're trying to focus on where AI can provide real tangible business value, not just cute or amusing results. We're focusing on apps built on top of the models while trying not to get to attached to any one model yet. The 💩 show over at OpenAI recently should give everyone pause about becoming too dependent on any one provider. We're also starting to think about the unique data sets we have in our business and new ones we can create that may produce value down the road. How about you? How are you thinking about the impact AI will have on your business?