March #TechBytes - 2nd part

Here is a roundup of some of the tech news you may have missed recently selected by us at Etiqa!

Ours is a journey inside the latest advanced technologies, news about the most exciting startups and scale ups from around the world, and updates on specialized, high-risk industries.

Stability AI CEO resigns advocating for decentralized AI

In a surprising move, Emad Mostaque, the CEO and founder of Stability AI , has stepped down from his position and the company's board. This departure marks the latest shakeup in the world of AI startups. Mostaque, known for his visionary leadership, emphasized the need for decentralized AI initiatives, suggesting that centralized AI might not be the ultimate solution.

Stability AI, backed by prominent investors like Lightspeed Venture Partners and Coatue Management, has yet to name a permanent replacement for the CEO role. However, the company has appointed Shan Shan Wong, the current COO, and Christian Laforte, the CTO, as interim co-CEOs.

Mostaque's decision to resign stems from his belief in the necessity of transparent and distributed governance in the AI sector. Expressing concerns about the concentration of power in the hands of a few, Mostaque advocates for a more inclusive approach to AI development and governance.

His departure comes at a pivotal time for Stability AI, which has faced challenges in talent retention and fundraising. Despite these hurdles, the company remains committed to its mission, particularly with its popular image generation tool, Stable Diffusion.

Google introduces Gemini AI to boost Gmail productivity in workspace

谷歌 's recent announcement regarding the integration of Gemini into its Workspace suite, particularly in Gmail, has been met with anticipation from the tech community. This move aligns with Google's broader strategy to accelerate advancements in artificial intelligence, following Microsoft's example with Copilot. The Gemini integration extends to various services within Workspace, promising enhanced productivity across platforms.

Under the new One AI Premium plan, users gain access to Gemini Advanced, Google's most sophisticated AI model. This integration not only enhances Gmail but also extends to Documents, Presentations, Sheets, and Meet. Users can execute specific tasks directly within these applications, saving time by bypassing the need to access Gemini's web page or app version.

Gemini aims to streamline workflow management by facilitating faster email composition and efficiently summarizing lengthy email threads, extracting crucial information buried within the inbox. Acting as an assistant, Gemini automates email composition, provides context-specific response suggestions, and highlights key details from email threads while posing specific questions.

Google emphasizes its commitment to user privacy and data protection, ensuring that data and prompts won't be used to train or improve generative AI models outside of Workspace without explicit authorization.

While additional features of Gemini AI are currently rolling out for English-language AI Premium subscribers, Italian-language support is forthcoming. In Italy, users can access Gemini in Gmail by using a VPN to change their IP address to the US, enabling access through an American server.

Meta's AI comes to WhatsApp: changes to the search bar

Artificial intelligence is increasingly being integrated into various platforms to streamline user activities, from content creation to finding useful information, fundamentally altering how people interact with their devices daily.

Meta has decided to leverage AI in its flagship messaging app, WhatsApp , aiming to enhance user experience and maintain competitiveness.

Meta's AI will be directly integrated into the search bar, making processes faster and eliminating the need to personally initiate AI chats. The benefits following this integration could potentially revolutionize how WhatsApp is utilized.

Meta's upcoming updates for WhatsApp include a new feature based on artificial intelligence directly integrated into the search bar. Currently in the testing phase, this change is expected to roll out soon to all WhatsApp users.

Specifically, a prompt labeled 'Ask Meta AI' has been placed below the search bar, a command that reduces steps to interact with Meta's artificial intelligence, facilitating smoother message exchanges.

Meta AI serves as a powerful and versatile virtual assistant, capable of providing quick, accurate, and relevant responses while aiding users in various tasks. The key advantage of this new functionality on WhatsApp is the ability to access the chatbot whenever needed, without manually initiating a chat with Meta's AI. Queries can be typed into the search bar or selected from suggested prompts.

Direct access to the AI virtual assistant from the search bar is just one of the upcoming changes concerning artificial intelligence on WhatsApp. Meta is heavily investing in this aspect not only in the messaging app but across all its platforms.

WhatsApp developers are reportedly working on additional implementations, such as high-level image editing and generation tools based on AI.

Google unveils SIMA: the new AI game player

DeepMind, acquired by Google in 2014, has introduced SIMA (Scalable Instructable Multiworld Agent), an AI designed to play video games. SIMA, still in its experimental phase, aims to perform a wide range of gaming tasks, allowing users to interact within the game environment.

The utility of an AI capable of gaming extends beyond entertainment.

DeepMind views video games as crucial testing grounds for AI systems due to their rich learning environments. While AI has been applied in gaming previously, SIMA represents a significant advancement as an AI agent trainable to play various games.

SIMA's development involved collaboration with eight game developers, including Embracer, Tuxedo Labs, Hello Games, and Coffee Stain. It has been trained on nine different games, such as Hello Games' No Man's Sky and Tuxedo Labs' Teardown.

Currently, SIMA can perform 600 basic abilities, including navigation, object interaction, and menu usage, within about 10 seconds. What sets SIMA apart is its versatility; it doesn't require access to game source code or custom APIs. Instead, it operates solely based on screen images and natural language instructions from users, controlling the game character through keyboard and mouse inputs.

This simplicity enables SIMA to potentially interact with any virtual environment, marking a significant step forward in AI's integration into gaming.