Good evening, in today’s AI news Elon Musk’s xAI raises $6B, Meta in the works of a paid AI assistant, and more.

Apple is planning to launch AI-emojis

Apple’s upcoming Worldwide Developers Conference (WWDC) is expected to unveil their plans for integrating Artificial Intelligence (AI) into their device. This may not involve the dramatic features showcased by competitors, but rather focus on practical applications.

One potential feature is a “smart recap” function that summarizes missed messages, notifications and other content. This could be particularly useful for managing iOS notifications. AI-generated transcripts for voice memos is another anticipated update, along with improved search functions and writing suggestions. 

There would also be AI powered improvements to on-device Spotlight search, internet searches and Safari. Siri might also get an update to sound more natural. Overall, these AI features are expected to prioritize on-device processing for faster performance. 

Meta in the works of a paid AI assistant

To compete with its contemporaries, Meta is planning to launch a paid version of its AI assistant—Meta AI. While information on the cost and features of this paid version are yet to be disclosed, it will use virtual and augmented reality to provide unique offerings. The news originally viewed on The Information also suggests that Meta is working on an AI coding tool, similar to Microsoft’s CoPilot.

Niantic uses Meta’s Llama

Niantic's new mobile AR game, Peridot, takes virtual pets to a new level with the help of generative AI. Niantic uses Meta’s Llama 2 to create responsive AR pets called Dots that react realistically to their surroundings and the player’s interactions. 

This allows for a more dynamic and unique gameplay experience, fostering deeper connections between players and their virtual companions. Leveraging Llama 2’s ability to “learn” and respond creatively, Niantic is able to deliver unpredictable and surprising behaviors from the Dots, making them feel more lifelike. This technology has the potential to revolutionize the way players interact with virtual creatures in the future. 

Google AI introduces AGREE to improve self-grounding in LLMs

A new framework called AGREE has been introduced by Google to address the issue of factual accuracy in large language models. AGREE enables LLMs to cite their sources and improve the grounding of their responses. 

It achieves this through a two-step process. First, LLMs are trained on data that includes citations, allowing them to generate responses within citation built-in. Second, during use, the LLM can search for additional information to support its initial claims, further enhancing grounding and reliability.

This method offers a significant improvement over existing techniques, leading to more trustworthy and verifiable LLM outputs. The research encourages further development in this area to create LLMs that are more reliable and beneficial for real-world applications. 

Google introduces LearnLM to assist people in learning

Google has introduced LearnLM, a new line of AI models designed to improve the way people learn. LearnLM is based on Google's existing Gemini model and incorporates principles from educational research to create a more engaging and effective learning experience.

Some examples of how LearnLM will be used include:

• In Google Search, users will be able to adjust the way information is presented to better suit their needs.

• On Android devices, Circle to Search will be expanded to include support for more complex math and physics problems.

• A new "Learning Coach" feature within Gemini will provide step-by-step guidance and practice activities.

• On YouTube, viewers will be able to ask clarifying questions and take quizzes directly within educational videos.

• Google Classroom will leverage LearnLM to help teachers discover new lesson ideas and materials.

In addition to these features, Google is developing new tools like "Illuminate" which can generate audio summaries of research papers, and "Learn About" which uses a combination of content, learning science, and chat experiences to guide users through a topic.

xAI raises $6B from a16z, Sequoia and others

Within one year of xAI’s inception, xAI has announced a series B round of $6 billion from a16z, Sequoia, and others. After the new funding, the company is valued at $26 billion. xAI will use the new funds to market, build advanced infrastructure, and accelerate the research and development of future technologies. 

That’s all for today’s AI and tech news, what do you think about it? Let us know in the comments.