Posted by Joseph Lewis – Technical Author, Android AI
AI on Android Highlight Week this yr runs September thirtieth to October 4th! As a part of the Android “Highlight Weeks” collection, this week’s content material and updates are your gateway to understanding tips on how to combine cutting-edge AI into your Android apps. Whether or not you are a seasoned Android developer, an AI fanatic, or simply beginning out in your improvement journey, prepare for every week full of insightful periods, sensible demos, and galvanizing success tales that’ll provide help to construct intuitive and highly effective AI integrations.
All through the week, we’ll dive into the core applied sciences driving AI experiences on Android. This weblog might be up to date all through the week with hyperlinks to new bulletins and assets, so examine again right here every day for updates!
Monday: Getting began with AI
September 30, 2024
Learn to start with AI on Android improvement. Perceive which AI fashions and variations you may work with. Find out about developer instruments that will help you begin constructing options empowered with AI.
We’ll information you thru the variations between conventional programming and machine studying, and distinction conventional machine studying with generative AI. The submit explains giant language fashions (LLMs), the transformer structure, and key ideas like context home windows and embeddings. It additionally touches on fine-tuning and the way forward for LLMs on Android.
Learn the weblog submit: A fast introduction to giant language fashions for Android Builders
We’ll then present a glance behind the scenes at our work bettering developer productiveness with Gemini in Android Studio. We’ll talk about Studio’s new AI code completion characteristic, how we have been working to enhance the accuracy and high quality of steered code, and the way this characteristic can profit your workflow.
Learn the weblog submit: Gemini in Android Studio: Code Completion features highly effective mannequin enhancements
Tuesday: On-device AI capabilities with Gemini Nano
October 1, 2024
Uncover how Gemini Nano empowers Android builders to unlock the complete potential of generative AI, providing personalization and privateness advantages for next-generation apps. We’ll share how one can start integrating your Android apps with on-device LLMs. Search for extra info and bulletins right here on Tuesday!
Wednesday: On-device AI with customized fashions
October 2, 2024
On Wednesday, we’ll provide help to perceive tips on how to deliver your personal AI mannequin to Android units, and how one can combine instruments and applied sciences from Google and different sources. The power to run subtle AI fashions straight on units – whether or not it is a smartphone, pill, or embedded system – opens up thrilling prospects for higher efficiency, privateness, usability, and value effectivity.
We’ll additionally provide you with an in depth walkthrough of how Android builders can leverage Google AI Edge Torch to transform PyTorch machine studying fashions for on-device execution, utilizing the LiteRT and MediaPipe Duties libraries. This walkthrough contains code examples and explanations for changing a MobileViT mannequin for picture classification and a DIS mannequin for segmentation, and highlights the steps concerned in making ready these fashions for seamless integration into Android purposes. By following this information, builders can harness PyTorch fashions to boost their Android apps with superior machine studying capabilities.
Thursday: Entry cloud fashions with Android SDKs
October 3, 2024
Faucet into the boundless potential of Gemini 1.5 Professional and Gemini 1.5 Flash, the revolutionary generative AI fashions which might be redefining the capabilities of Android apps. With Gemini 1.5 Professional and 1.5 Flash, you may have the instruments it’s worthwhile to create apps which might be actually clever and interactive.
On Thursday, we’ll provide you with a codelab that’ll provide help to perceive tips on how to combine the Gemini API capabilities into your Android initiatives. We’ll information you thru crafting efficient prompts and combine Vertex AI in Firebase. By the top of this hands-on tutorial, you can implement options like textual content summarization in your personal app all powered by the cutting-edge Gemini fashions.
Subsequent we’ll publish a weblog submit exploring the potential of the Gemini API with case research. We’ll delve into how Android builders are leveraging generative AI capabilities in modern methods, showcasing real-world examples of apps which have efficiently built-in the Gemini API. From meal planning to journaling and personalised consumer experiences, the article highlights examples of how Android builders are already benefiting from Gemini transformative capabilities of their apps.
We’ll additionally share with you examples of superior options of the Gemini API to transcend easy textual content prompting. You will find out how system directions can form the mannequin conduct, how JSON help streamlines improvement, and the way multimodal capabilities and performance calling can unlock thrilling new use instances in your apps.
Friday: Construct with AI on Android and past
October 4, 2024
Because the capstone for AI on Android Highlight Week, we’ll host a dialogue with Kateryna Semenova, Oli Gaymond, Miguel Ramos, and Khanh LeViet to speak about constructing with AI on Android. We’ll discover the most recent AI developments tailor-made for Android engineers, showcasing how these applied sciences can elevate your app improvement sport. By participating discussions and real-world examples, we’ll unveil the potential of AI, from quick, non-public on-device options utilizing Gemini Nano to the highly effective capabilities of Gemini 1.5 Flash and Professional. We’ll talk about constructing generative AI options quickly utilizing Vertex AI in Firebase. And we’ll dive into harnessing the facility of AI with security and privateness in thoughts.
Work with Gemini past Android
As we wrap issues up for AI on Android Highlight Week, know that we’re striving to supply complete AI options for cross-platform Gemini improvement. The AI capabilities showcased throughout Android AI Week can prolong to different platforms, reminiscent of built-in AI in Chrome. Internet builders can leverage related instruments and strategies to create internet experiences enhanced by AI. Builders can run Gemini Professional within the cloud for pure language processing and different complicated consumer journeys. Or, you discover the advantages of performing AI inferenceclient-side, with Gemini Nano in Chrome.
Construct with usability and privateness in thoughts
As you embark in your AI improvement journey, we would like you to bear in mind just a few necessary concerns:
- Privateness: Prioritize consumer privateness and knowledge safety when implementing AI options, particularly when dealing with delicate consumer info. When it turns into out there, go for on-device AI options like Gemini Nano each time attainable to attenuate knowledge publicity.
- Seamless consumer expertise: Be certain that AI options seamlessly combine into your app’s general consumer expertise. AI ought to improve the consumer expertise, not disrupt it.
- Moral concerns: AI applied sciences are developed and deployed in a means that advantages society whereas minimizing potential hurt. By contemplating equity, transparency, privateness, accountability, and societal impression, builders can play an important function in making a future the place AI serves humanity’s finest pursuits. Be aware of the moral implications of AI, reminiscent of potential biases in your AI fashions. Try to create AI-powered options which might be truthful and inclusive.
AI on Android Highlight Week is a chance to discover the most recent in AI and its potential for Android app improvement. We encourage you to delve into the wealth of assets shared in the course of the week and start experimenting with AI in your personal initiatives. The way forward for Android is rooted in AI and machine studying, and with the instruments and data shared throughout Android AI Week, builders are well-equipped to construct the subsequent era of AI-powered apps.
What’s subsequent
Come again to this weblog submit for updates; we’ll add hyperlinks to weblog and video content material and extra all through the week. Comply with Android Builders on X and Android Builders at LinkedIn, and keep in mind to make use of the hashtag #AndroidAI to share your AI-powered Android creations, and be a part of the colourful group of builders pushing the boundaries of cellular AI.
