Google has been facing fierce competition from other tech giants in the field of artificial intelligence (AI), especially in natural language processing (NLP) and generative AI. Chat-GPT, the large language model developed by OpenAI, has been widely adopted by many companies and platforms for various tasks such as chatbots, content creation, and text summarization. Microsoft has also been investing heavily in AI research and development, launching its own large language model called Turing-NLG and acquiring Nuance, a leader in speech recognition and conversational AI.
To accelerate its growth and hopefully bridge the innovation gap, Google unveiled new AI and software at Google I/O 2023. Some of the highlights included a much-needed new and improved Bard, a brand new large language model, and a generative AI developer interface. Here’s a roundup of what Google announced and how you can take advantage.
Bard 2.0: A more powerful and versatile generative AI tool
Bard is Google’s generative AI tool that allows users to create various types of content such as poems, stories, essays, code, lyrics, and more. It was launched in 2021 as an experimental feature in Google Docs, but it soon gained popularity among users who wanted to unleash their creativity with the help of AI.
At I/O 2023, Google announced Bard 2.0, a major upgrade that brings several new features and improvements to the tool. Bard 2.0 is powered by PaLM 2, Google’s new AI language model that underpins more than two dozen of its products, including Gmail, Search, Maps, and Translate. PaLM 2 is faster and more efficient than its predecessor, and can run locally on mobile devices.
Bard 2.0 also introduces a new feature called “Help me write”, which allows users to generate full drafts of texts based on simple prompts or keywords. For example, if you want to write an email asking for a refund from an airline, you can just type in “an email that asks for a full refund” and hit create. Bard 2.0 will generate a full draft that pulls in relevant details from previous emails or other sources. You can then edit or refine the draft as you wish.
Another new feature is “Bard for Developers”, which is a generative AI developer interface that enables developers to create custom applications using Bard 2.0’s capabilities. For example, you can use Bard for Developers to create a chatbot that can answer questions about your product or service, or a content generator that can produce blog posts or social media posts for your brand.
Bard 2.0 is available as part of Google Workspace updates, and you can also access it through Google Labs, where you can find other experimental applications powered by generative AI.
PaLM 2: A new large language model that runs on mobile devices
PaLM 2 is Google’s new large language model that was unveiled at I/O 2023. It is a successor to PaLM, which was launched in 2019 as a general-purpose language model that could perform various NLP tasks such as question answering, text summarization, sentiment analysis, and more.
PaLM 2 is based on the Transformer architecture, which is the same one used by Chat-GPT and Turing-NLG. However, PaLM 2 has some unique features that make it stand out from its competitors. First of all, PaLM 2 is much smaller and lighter than Chat-GPT and Turing-NLG, which have billions of parameters and require massive computing power to run. PaLM 2 has only about 100 million parameters and can run locally on mobile devices without relying on cloud servers.
Secondly, PaLM 2 is more adaptable and versatile than Chat-GPT and Turing-NLG, which are trained on large corpora of text from the internet and may not be suitable for specific domains or tasks. PaLM 2 can be fine-tuned on smaller datasets or user feedback to perform better on different domains or tasks. For example, PaLM 2 can be fine-tuned on medical texts to create Med-PaLM, a specialized language model that could be a helpful tool for clinicians.
PaLM 2 is already powering more than two dozen of Google’s products, such as Gmail, Search, Maps, Translate, Lens, Photos.
Search Generative Experience: A new way to explore and learn from the web
Search is Google’s core product and one of the most widely used services in the world. It helps people find information, answers, and solutions from billions of web pages. But sometimes, people want more than just a list of links. They want to dig deeper into a topic, learn new things, or discover new perspectives.
That’s why Google introduced the Search Generative Experience (SGE) at I/O 2023. SGE is a new feature that uses generative AI to create interactive and immersive snapshots that help users explore and learn from the web. SGE is part of the new experimental Search Labs platform, where users can sign up to test new projects and provide feedback.
For example, let’s say you want to learn more about quantum computing. You can type in “quantum computing” and hit SGE. You’ll see a snapshot that summarizes the main concepts, applications, and challenges of quantum computing. You can also see related topics, such as quantum cryptography, quantum algorithms, or quantum supremacy. You can click on any of them to see another snapshot that explains them in more detail.
But SGE is not just a passive experience. You can also interact with the snapshots, ask questions, or request more information. For example, you can ask “How does quantum cryptography work?” and SGE will generate a brief answer based on the web sources. You can also request more information by typing in “Show me more” or “Tell me more”. SGE will generate additional content, such as examples, diagrams, or videos.
SGE is not only useful for learning, but also for exploring your interests and passions. For example, if you’re a fan of music, you can type in “music genres” and hit SGE. You’ll see a snapshot that shows you different types of music genres, such as rock, jazz, hip hop, or classical. You can click on any of them to see another snapshot that gives you more information about their history, characteristics, and artists. You can also listen to samples of songs from each genre, or discover new genres that you might like.
SGE is available for three topics at launch: quantum computing, music genres, and climate change. Google plans to add more topics and features in the future based on user feedback and demand.
Conclusion
Google I/O 2023 was a showcase of how Google is using generative AI to make its products more helpful for everyone. From Bard 2.0 to PaLM 2 to SGE, Google demonstrated its bold and responsible approach to AI innovation. Whether you’re a writer, a developer, a learner, or an explorer, Google has something for you to try out with generative AI.
Watch the full recap:
Share to your social below!
