February 2024 Summaries

Google Cloud Next 2024 is set to take place from April 9-11 at the Mandalay Bay Convention Center in Las Vegas, offering a wide array of sessions and activities catering to developers, data scientists, and businesses of all sizes. The event promises an immersive experience with interactive demos, talks from Champion Innovators, and numerous opportunities for networking and learning. Featuring a significant increase in advanced technical sessions, the event covers diverse topics including big data, AI model building using Vertex AI, security across various cloud platforms, and discussions on diversity, equity, and inclusion. Attendees can expect to engage with thousands of Googlers, participate in specialized interest groups, and explore cutting-edge technologies, making it an essential gathering for those interested in cloud technology and innovation.

Gemma is a new family of lightweight, state-of-the-art open models available in the KerasNLP collection, compatible with JAX, PyTorch, and TensorFlow, and designed specifically for large language models. These models, offered in 2B and 7B parameter sizes, outperform similar models and some larger ones on benchmarks like MMLU, GSM8K, and HumanEval. Keras 3 introduces new features like a LoRA API for parameter-efficient fine-tuning and large-scale model-parallel training, allowing Gemma models to be fine-tuned with fewer parameters. The models can be instantiated easily with a single line of code and benefit from Keras's new distribution API for distributed training, particularly on the JAX backend due to its scalability. Users can explore various training setups, such as using TPUv3 or multiple GPUs, and are encouraged to share their fine-tuned models on platforms like Kaggle.

Gemini 1.0 Ultra has been released in Gemini Advanced, and Google is now introducing Gemini 1.5, which employs a Mixture-of-Experts approach to enhance efficiency by routing requests to smaller expert neural networks for faster and higher-quality responses. Available for Private Preview, Gemini 1.5 Pro, a mid-sized multimodal model, features an experimental 1 million token context window, significantly surpassing the previous largest context window of 200,000 tokens. This expanded window allows developers to upload large volumes of data, such as PDFs, code repositories, and videos, into Google AI Studio and enables deeper analysis and multimodal reasoning. The platform, available in 38 languages across over 180 countries, facilitates easy integration of the Gemini API into applications, with new developer tools and extensions for further customization. Additionally, pricing for Gemini 1.0 Pro has been reduced, and pay-as-you-go plans are forthcoming, as Google continues to refine and roll out these advanced AI capabilities to developers worldwide.

Project IDX has introduced new AI-powered features using Gemini models to enhance the developer workflow for full-stack, multiplatform software development. These features include AI-driven code completion, an assistive chatbot, and contextual code actions, all integrated into a centralized workspace. Developers can leverage inline AI assistance for real-time error correction and code suggestions, enabling a more efficient and higher-quality coding process. The platform also offers a Gemini API template that simplifies embedding AI features into applications, supporting various input modalities and context-aware outputs. This template allows developers to create applications with functionalities like conversational interfaces and automatic image captioning. Additionally, Project IDX plans to expand AI availability to over 150 countries and encourages user feedback to continue improving the platform.

Environmental science student Caitlin and French developer Nicolas collaborated to create the app "We Spot Turtles!" after being inspired by their experiences with sea turtle conservation in Western Australia and French Polynesia. The app allows users to upload photos of sea turtles, which are then automatically geolocated and timestamped to aid in tracking these critically endangered species. Utilizing artificial intelligence, the app identifies individual turtles by analyzing their facial scales and pigmentations, assigning unique codes for better monitoring. By launching on Google Play, the app has reached a global audience, fostering communities in places like Exmouth, Western Australia, and Manly Beach, Sydney. Caitlin and Nicolas aim to empower public participation in conservation efforts and plan to expand their work to include other species such as whale sharks, birds, and red pandas, striving to enhance global conservation initiatives.

Google Pay has announced the general availability of liability shift for Visa device tokens, a feature that transfers the responsibility for losses from fraudulent transactions from merchants to issuing banks. Previously, only Mastercard device token transactions had such a shift, with Visa transactions limited to eligible device tokens from European issuing banks. To identify if a Visa transaction is eligible for liability shift, merchants and PSPs can check the eciIndicator value, which should be 05. Merchants in Europe are automatically enrolled, while others must opt-in via the Google Pay & Wallet console. However, not all transactions qualify for liability shift, as certain parameters and conditions must be met, such as matching transaction currencies and specific merchant category codes (MCC) exclusions in the US. Google Pay provides resources, including documentation and a sample application, to help merchants integrate the service and ensure their transactions benefit from liability shift.

Google Developer Student Clubs (GDSCs) are university-based community groups aimed at students interested in learning and applying Google technologies through hands-on projects and events. With over 2000 chapters in more than 100 countries, these clubs offer opportunities to delve into technologies like AI, Machine Learning, and Google Cloud, while fostering skills in Android development and Flutter. GDSC Leads, who manage these clubs, receive mentorship from Google and connect with a global network of leaders, enhancing their technical and leadership abilities. Events such as hackathons and Solution Challenges provide platforms for members to address real-world issues using technology. Notable initiatives include the DevFest On Campus in Canada, organized by Drashtant Chudasama, and the "Unlocking the Power of Blockchain" event in India led by Amitasha Verma, which showcase the clubs' impact in promoting tech skills and innovation. Students interested in leading a GDSC can apply to become a GDSC Lead, gaining valuable experience in community management and technological problem-solving.

On July 22, 2024, the Privacy Sandbox initiative announced a shift in focus towards enhancing user choice rather than eliminating third-party cookies, while maintaining investment in the Federated Credential Management (FedCM) API through Google Identity Services (GIS). The transition to FedCM is part of Chrome's move to phase out third-party cookies, and GIS developers will be automatically migrated to the FedCM API starting in April 2024. Most developers will experience a seamless transition through backwards-compatible updates, though some with custom integrations may need to make minor changes. The migration involves a timeline where Chrome began testing third-party cookie restrictions at 1% in January 2024, with full enforcement expected once remaining concerns are addressed by the Competition and Markets Authority (CMA). Developers can temporarily exempt their traffic from using FedCM until February 2025 if needed. The FedCM API aims to provide a privacy-preserving alternative for sign-in experiences, ensuring a secure and streamlined process for websites using GIS's One Tap and Automatic Sign-In features. Developers are encouraged to prepare for the migration by testing FedCM functionality and making necessary changes to ensure continuity in user flows.

People of AI returns for its third season, featuring a diverse lineup of guests from the world of Artificial Intelligence, co-hosted by Luiz Gustavo Martins. This season delves into the personal and professional journeys of individuals in AI, focusing on their work and its societal implications. Notable guests include Adrit Rao, a high school student creating healthcare apps; Indira Negi from the Bill and Melinda Gates Foundation discussing AI investments in healthcare; Tris Warkentin and Kathleen Kenealy from Google DeepMind exploring new AI models; and François Chollet, creator of Keras, discussing AI innovations at Google. The podcast, sponsored by Google, aims to align technology with public narratives by exploring the potential and challenges of AI, emphasizing the importance of responsible use. Each episode alternates between video and audio, with new content released weekly.

At CES, Google introduced a 360° event-scale augmented reality (AR) experience at their booth to enhance attendee engagement, utilizing Google’s latest AR technologies in partnership with Left Field Labs and Adobe. The experience featured the Android Bot as a virtual guide, offering seamless navigation and delightful AR content, such as generative AI wallpapers and interactive versions of Android on Sphere. Geospatial Creator, powered by ARCore and Photorealistic 3D Tiles from Google Maps Platform, was pivotal in visualizing and anchoring digital content within the real-world setting of the Las Vegas Convention Center. Attendees accessed the experience through a simple QR scan, bypassing the need for app downloads, and could interact with various demo stations and navigation tools. The initiative showcased how location-based AR can transform event experiences by blending digital and physical worlds, offering both entertainment and practical navigation solutions.

Sarah Pinner's commitment to reducing waste, sparked in childhood, led her to co-found Beni, an AI-powered browser extension that connects shoppers with resale options to promote the circular economy and lessen the environmental impact of fashion. Beni, co-founded with CTO Celine Lightfoot, aggregates data from over 40 marketplaces, enabling users to find secondhand apparel while shopping online, typically saving 55% compared to new items. The platform leverages AI to streamline data from partners and normalize product sizing, enhancing user experience. Beni was an early player in the recommerce platform arena, and its participation in the Google for Startups Accelerator: Circular Economy program has been instrumental in refining its technology and growth strategies. This collaboration provided Beni with valuable insights into using Google tools effectively, improving search relevance, and fostering connections with other circular economy businesses. The mentorship received, particularly in areas like UX writing for sustainability and transitioning from Vision API Product Search to their own reverse image search, was pivotal in Beni's development. Beni continues to expand its platform capabilities and aims to integrate resale into broader e-commerce platforms, striving to normalize secondhand shopping.

Carbon Limit, a company based in Boca Raton, is dedicated to decarbonizing the industry by creating sustainable solutions for the built environment, exemplified by their flagship product CaptureCrete, which captures and stores CO2 from the air within concrete. The company's founders, driven by personal connections to their mission, have developed a strong culture and have pivoted their technology to focus on generating valuable carbon credits. Embracing artificial intelligence, Carbon Limit has enhanced transparency and efficiency in carbon credit verification through digital tools and AI-powered data analysis. Utilizing Google technology and participating in the Google for Startups Accelerator: Climate Change program, Carbon Limit has integrated AI and machine learning into their processes, improved their SEO, and gained mentorship and resources that have significantly contributed to their growth. The program also facilitated a partnership with Google, boosting Carbon Limit's credibility. Looking forward, the company plans to launch technology for data centers to mitigate electricity use and reduce CO2 pollution, aiming to inspire broader environmental awareness and action.