Metis may launch as soon as September.
Amazon is reportedly developing a new AI chatbot named “Metis,” according to sources familiar with the project and an internal document obtained. This next-generation chatbot is poised to leverage a novel AI model known as Olympus, marking a strategic pivot from Amazon’s existing Titan model. The core innovation in Metis lies in its use of Retrieval-Augmented Generation (RAG), a sophisticated technique that allows the AI to pull data from external, authoritative sources in real-time. This approach not only enhances the chatbot’s response accuracy by accessing the most current information but also offers greater transparency on how responses are formulated, providing users a peek into the AI’s decision-making process.
Under the leadership of Amazon CEO Andy Jassy and Rohit Prasad, the senior vice president and head scientist of Amazon’s AGI team, Metis is being crafted by a skilled team, including many veterans from the Alexa AI division. The technology underpinning Metis borrows heavily from advancements made for Amazon’s upcoming voice assistant, “Remarkable Alexa.” Amazon aims to unveil Metis in September, aligning with the company’s annual product launch event. Despite the technical readiness of Metis, there are concerns about its market entry timing. Competitors like OpenAI’s ChatGPT, launched in November 2022, and Google’s Bard (renamed Gemini) released in March 2023, have already established a presence in the market. Amazon’s previous AI, Titan, has struggled to match the capabilities of these industry leaders, although Amazon has attempted to broaden its reach with variants tailored for business applications.
Metis represents a significant step for Amazon in the competitive landscape of AI chatbots, as it seeks to integrate cutting-edge retrieval technology with its rich history in AI to deliver a more dynamic and informed user experience.
Google is reportedly developing AI chatbots modelled after celebrities and influencers.
Chatbots modelled after celebrities? This innovative approach could redefine interactions in the digital space.
Google is reportedly developing AI-powered chatbots that emulate celebrities and YouTube influencers, leveraging its advanced Gemini language models. This venture, reported by The Information, aims to enrich Google’s AI offerings by incorporating familiar public figures into its chatbot technology. While similar concepts exist—such as Meta’s celebrity chatbots and Character.ai’s personality-based bots—Google’s version promises a unique twist by allowing users to create custom chatbots by specifying desired traits and appearances. The initiative is part of Google’s broader strategy to stay competitive in the dynamic AI landscape, spearheaded by Ryan Germick from the Google Doodles team, along with a dedicated ten-member squad. Although primarily an experimental venture that may initially appear on Google Labs, Google’s approach involves forming partnerships with high-profile personalities, potentially offering a more engaging way for fans to interact with their favourite celebrities digitally. Meta’s similar projects, like chatbots mimicking TikTok star Charli D’Amelio or rapper Snoop Dogg, have seen mixed success, suggesting that the appeal of celebrity-based chatbots might be limited. However, Google’s implementation could differentiate itself by tapping into the company’s robust AI research and broad user base. This project not only reflects the ongoing evolution of AI in creating personalised digital experiences but also highlights the tech industry’s shift towards more interactive and customizable user engagement platforms.SOURCE
Figma announces big redesign with AI
Figma is set to enhance its design platform by introducing a new presentation feature, expanding its capabilities beyond traditional design tools to include dynamic presentation options directly within the application. This update will allow users to seamlessly create and share presentations, further streamlining the workflow for designers and teams.
Figma announced several new features at its Config conference today, including a comprehensive UI redesign, the introduction of generative AI tools, and integrated slideshow capabilities. The redesign aims to set the stage for Figma’s development over the next ten years. Key updates include a revamped toolbar, smoother rounded corners, and the addition of over 200 new icons. These changes are part of Figma’s strategy to minimise the prominence of its user interface on the canvas, thereby spotlighting user projects more effectively. The update is designed to be user-friendly for newcomers while retaining robust functionality for seasoned Figma users. This balance aims to enhance the overall user experience, making the platform more intuitive and adaptable across different skill levels.
Figma has introduced its third major redesign since its inception during the closed beta phase, signalling continued evolution in its user interface. This latest redesign is currently available through a limited beta, with interested users encouraged to join a waitlist to explore the new features firsthand. The centrepiece of Figmas latest update is the introduction of generative AI tools, designed to streamline the design process. These tools represent a specialised adaptation of the AI functionalities that have become common in applications such as email drafting. Figma’s implementation focuses on enhancing design efficiency, enabling users to quickly generate preliminary designs directly within the platform. During a product briefing, Figma’s Chief Product Officer, Yuhki Yamashita, demonstrated the capabilities of these new tools. He showcased how quickly the AI could generate a functional app design for a restaurant, complete with menu options, navigation tabs, and integration buttons for third-party delivery services like Uber Eats and DoorDash. The initial output resembled a standard mobile app interface, which Yamashita then immediately began customising.
Further illustrating the utility of the AI tools, Yamashita used Figma AI to create a recipe page layout for chocolate chip cookies. The AI promptly produced not only the layout but also an AI-generated image of a cookie, demonstrating the tool’s ability to handle both structural design elements and graphic content creation. This feature underscores Figma’s commitment to leveraging AI to facilitate rapid design prototyping, offering significant time savings and increased creative flexibility for designers. Figma is enhancing its design platform by integrating new AI-driven features designed to streamline and accelerate various design tasks. These innovations include an “AI-enhanced” asset search that simplifies the process of finding specific design elements, and the ability to auto-generate contextual text within designs, moving away from the traditional use of generic Lorem ipsum text.
These AI tools are crafted to democratize design proficiency, making it easier for beginners to experiment and develop ideas, while also enabling seasoned users to refine and iterate designs more efficiently. Yuhki Yamashita, Figma’s Chief Product Officer, described the aim of these tools as “using AI to lower the floor and raise the ceiling” for Figma users, a sentiment echoed by Figma CEO Dylan Field in discussions about the platform’s direction.
The rollout of these AI capabilities will commence with a limited beta starting Wednesday, for which users can sign up to join the waitlist. Figma plans to run this beta phase through the end of the year, during which the AI tools will be available at no cost. However, Figma has indicated that it might implement usage limits during this period and will provide clear pricing details ahead of the official launch of these features. This approach ensures that users can familiarise themselves with the functionalities and potential of the AI tools while also preparing for any transition to paid features.
Figma has provided detailed information about the ethical and technical foundations of its AI model training practices in a recent blog post by Kris Rasmussen, the company’s CTO. In the post, Rasmussen emphasises that the new generative features introduced by Figma rely on third-party AI models and have not been trained on any private files or customer data from Figma users. This clarification comes as part of Figma’s commitment to user privacy and transparency, particularly in light of the industry’s heightened sensitivity to data usage in AI training, exemplified by recent controversies like Adobe’s terms of service misunderstanding. To enhance the functionality of its AI tools, Figma has fine-tuned its visual and asset search capabilities using images of user interfaces from publicly available files in the Figma Community, which are free to access. This approach allows the AI to learn from a broad dataset while ensuring that no private user data is compromised. The models are designed to recognize patterns and understand Figma-specific tools and concepts without relying on the proprietary content of individual users. Moreover, Figma is introducing controls that allow administrators within the platform to decide if their organisation’s content can be used to train Figma’s AI models. This content includes any material created or uploaded by users, such as layer names, properties, text, images, comments, and annotations. Starting August 15th, user consent will be required for this training process, with default settings varying by plan: Starter and Professional plans will automatically opt-in users, whereas Organization and Enterprise plans will not, giving larger entities more control over their data. Alongside these AI developments, Figma is launching Figma Slides, a new tool designed to facilitate the creation and sharing of presentations directly within the Figma platform. This feature responds to user behavior, as many have been creatively using Figma for slideshow presentations. Figma Slides aims to streamline this process, providing an official and integrated solution for users to efficiently craft and distribute their presentations without resorting to workarounds.
These updates signify Figma’s ongoing efforts to balance innovation with user privacy and control, reflecting its strategic response to the evolving needs of its user base and the broader tech community’s expectations around AI and data security.
Figma is introducing several innovative features tailored specifically for designers as part of its latest updates, which include Figma Slides—a new tool that significantly enhances presentation capabilities directly within the Figma platform. One of the standout functionalities of Figma Slides is the ability for users to make real-time adjustments to designs within their presentations. Currently, any modifications made during a presentation won’t sync back to the original design files, but Figma plans to develop this capability in the future, aiming to streamline workflow and enhance design flexibility.
In addition to real-time editing, Figma Slides allows users to seamlessly present app prototypes from within the deck itself. This feature eliminates the need for cumbersome screen recordings often used to demonstrate how different parts of an app interact, providing a smoother and more integrated presentation experience. Furthermore, Figma Slides includes interactive elements such as polls or alignment scales, where audience members can actively participate by expressing their opinions or preferences during a presentation. Figma Slides will enter an open beta phase starting Wednesday, during which it will be available for free. Post-beta, it will transition to a paid feature, reflecting its value-add for professional design workflows. Additionally, Figma is enhancing its developer mode with new features like a “ready for dev” task list, which organises tasks for development teams and facilitates smoother transitions from design to development phases. These updates come in the wake of Adobe’s terminated $20 billion acquisition of Figma, which was called off due to regulatory pressures, resulting in Adobe paying a $1 billion breakup fee to Figma. This year’s Config conference marks the first since the cancellation of the acquisition, signalling a new phase of independence and innovation for Figma as it continues to expand its offerings and refine its platform for a growing user base of designers and developers. SOURCE
Ilya Sutskever Launches New AI Project
In a significant development in the artificial intelligence sector, Ilya Sutskever, a prominent figure in AI, has unveiled his latest venture, Safe Superintelligence (SSI). This new company is the first lab dedicated exclusively to the development of superintelligence, aiming to create AI that is not only advanced but also safe.
Sutskever, a co-founder of OpenAI, recently made headlines when he departed from the organisation amid rumours of internal disagreements regarding the focus and funding of AI safety and alignment projects. His departure followed a notable incident where he dismissed Sam Altman from the CEO position at OpenAI. Since leaving, Sutskever has been relatively quiet, sparking curiosity and speculation in the AI community. SSI marks a fresh start for Sutskever, who is joined by former OpenAI colleagues Daniel Levy and Daniel Gross in this endeavour. The new company is positioning itself distinctly from OpenAI by committing to prioritise long-term safety over immediate commercial success. This approach reflects a deliberate shift to avoid the “management overhead and short-term commercial pressures” that Sutskever felt hindered his previous efforts.
Despite the absence of a concrete business model, SSI is expected to attract substantial investment due to Sutskever’s influential track record in the AI industry, underscored by his pivotal role in the successes of OpenAI. His reputation as a visionary in the field has even drawn interest from tech mogul Elon Musk, among others.
SSI’s stated mission to develop a safe superintelligence reflects a growing recognition of the potential risks associated with AI. By focusing solely on this goal, SSI aims to lead the way in creating AI technologies that are beneficial and secure, ensuring that advancements in AI are aligned with human values and safety standards.
Conclusion:
The latest announcements from Amazon, Google, and Figma highlight pivotal advancements in AI technology. Amazon is enhancing its AI capabilities with the new Metis chatbot, aiming for a September launch. Google explores interactive AI chatbots modeled after celebrities to deepen user engagement. Meanwhile, Figma is broadening its design tools with innovative AI features and new presentation capabilities, demonstrating the rapid evolution of AI applications across different sectors. Each initiative showcases a strategic push towards integrating advanced technologies to improve user experience and functionality, reflecting the dynamic progress in the tech industry. Explore the Future with Arcot Group: Dive into a world of innovation and cutting-edge technology. Click here to discover how Arcot Group is leading the way in transformative solutions. Join us on our journey to excellence!
