The artificial intelligence (AI) realm witnesses rapid developments around the world, reshaping the world of digital transformation and accelerating the pace of AI adoption. In this weekly newsletter, FollowICT is highlighting the most important news and advancements in AI around the globe.
Microsoft is Developing AI Inference Models to Compete with OpenAI
Microsoft is developing AI inference models internally to compete with OpenAI, which it may sell to developers, according to The Information.
Microsoft has been looking to reduce its reliance on ChatGPT, even though its early partnership with the startup positioned it as a leader among major tech companies in the profitable AI race.
Microsoft Unveils Dragon Copilot, an Intelligent Healthcare Assistant
Microsoft has announced “Dragon Copilot,” an AI healthcare system that can, among other things, listen to patients’ complaints during their visits to clinics or hospitals and generate notes based on them.
The system combines speech recognition and ambient listening technologies developed by Nuance, a conversational AI company that Microsoft acquired in 2021.
The new system can help users streamline the documentation process of what happens in their surroundings and provide searches for general medical information from reliable content sources.
The intelligent assistant also has the capability to automate tasks; including verbal commands, note summaries, referral messages, and post-visit summaries.
Robots and Humans to Compete Against Each Other in the 1st Human-Robot Half Marathon
On 13 April, humanoid robots are set to compete in the first-ever half marathon featuring humans and humanoid robots racing against each other in Beijing.
Participating robots will have about three and a half hours to complete the course. The humanoid robots will compete alongside 12,000 human runners, with the top three winners—whether human or robot—receiving prizes, according to Xinhua News Agency
Moreover, the competing robots could win awards in three categories: race completion, best endurance, and most popular robot. The participating robots will run on a separate track.
Meta’s Upcoming Llama Models may Feature Advanced Voice Capabilities
Meta’s next major open AI model may focus on voice, according to a report in the Financial Times.
Meta is reportedly planning to introduce enhanced voice features with Llama 4, the next flagship in the Llama model family, expected to arrive in “weeks.” It is said that Meta is particularly focused on allowing users to interrupt the model while it is speaking, similar to OpenAI’s voice mode for ChatGPT and Google’s Gemini Live experience.
In comments this week at a Morgan Stanley conference, Chris Cox, Meta’s chief product officer, stated that Llama 4 will be a “versatile” model capable of interpreting and outputting both speech and other types of data.
X Allows Users to Direct Questions to Grok by Mentioning It in Comments
The X platform (formerly Twitter) is expanding access to its chatbot Grok to more users.
Users noticed on Friday that they can now mention “Grok” in comments on a post and ask a question in the comment to get an explanation about the post.
Typically, users of the X platform, owned by billionaire Elon Musk, access Grok through a button in the bottom bar of the platform interface.
Intangible secures $4 MLN for No-Code 3D Creation Tools for Filmmakers and Game Designers
Intangible, now backed by $4 million in seed funding, offers an AI-powered creative tool that allows users to create 3D world concepts with text prompts to assist creative professionals across various industries.
The company’s mission is to make the creative process accessible to everyone, including professionals such as filmmakers, game designers, event organizers, and marketing agencies, as well as everyday users looking to visualize concepts. For example, regular users can create home designs and small art projects using the tool.
With the new funding, Intangible plans to launch a no-code online 3D studio in June, it announced.
New Gemini Model for More Personalized Responses Using Browsing History
Google is preparing a new model of its smart assistant “Gemini” that can provide more personalized answers by giving the chatbot access to the user’s Google browsing history.
The upcoming model, called “Gemini Personalization,” can assist users based on their browsing history on Google.
The chat window for “Gemini Personalization” confirms that only the model will have access to the browsing history, and the conversation screen for this Gemini model includes a disclaimer stating that the model uses the user’s Google browsing history.
GibberLink App Enables AI Agents to Communicate With Each Other in Robot Language
A project that allows AI agents to talk on the phone to each other in a robotic language—unintelligible to humans—has gone viral on social media over the past week.
The project, called GibberLink, was created by two software engineers at Meta during a hackathon competition in London, hosted by ElevenLabs and Andreessen Horowitz.
The founders, Boris Starkov and Anton Bedkuyko, told TechCrunch in an interview that GibberLink allows an AI agent to recognize when it is on the phone with another AI agent. Once an AI agent realizes it is speaking to another AI agent, GibberLink prompts the agents to switch to a more efficient communication protocol called GGWave.
GGWave is an open-source library for sounds where each sound represents a small piece of data. This allows computers to communicate faster and more efficiently than they can with human speech. However, to the human ear, GGWave sounds like a series of “pings” and “pops”—exactly how you might imagine a computer’s native language sounds.
Apple Delays AI Developments for Siri Until 2026
Apple announced on Friday that it will delay the rollout of features that would enhance Siri’s ability to take actions within other apps until next year. This feature was expected to launch this spring.
Among other improvements for Siri that have been delayed was the ability to leverage what Apple calls “personal context.” For example, Siri would be able to fill out forms on behalf of users using their personal information taken from their driver’s license numbers from a photo.
An Apple representative stated:”We have also been working on developing a more personalized Siri, giving it more awareness of your personal context, as well as the ability to take actions on your behalf within and across your apps. It will take longer than we thought to roll out these features, and we expect to launch them next year”.
OpenAI and Oracle Seek Billions in Nvidia Chips For Stargate
OpenAI and Oracle plan to start outfitting a massive new data center in Texas with tens of thousands of powerful AI chips from Nvidia in the coming months, as part of efforts to build the first facility for their Stargate AI infrastructure project.
The site, located in the small city of Abilene, is expected to host 64,000 of Nvidia’s coveted “GB200” chips by the end of 2026, according to a Bloomberg report citing a source familiar with the matter.
The source indicated that the chips will be added to several halls in the data center in phases, with an initial installation of 16,000 expected to be completed by this summer.
The planned shipments represent a massive amount of computing power for just the initial stages of one data center for a single client.
This also underscores the potential scale of the joint Stargate project, which was revealed by OpenAI, SoftBank, and Oracle at an event held at the White House in January. OpenAI previously stated that Stargate would expand to up to ten locations.
Launch of NextGenAI Coalition to Boost Research and Education with $50 MLN in Investments
OpenAI announced the launch of NextGenAI, the first of its kind coalition comprising 15 leading research institutions dedicated to using AI to accelerate research achievements and transform education.
AI has the potential to drive progress in research and education, but only when people have the right tools to harness it. That’s why OpenAI is committed to allocating $50 million for research grants, computing funding, and access to APIs to support students, teachers, and researchers advancing the frontiers of knowledge.
By combining institutions across the United States and beyond, the NextGenAI initiative aims to stimulate progress at a faster rate than any institution could achieve alone. This initiative is not only designed to fuel the next generation of discoveries but also to prepare the next generation to shape the future of AI.
Founding partners of NextGenAI include the California Institute of Technology, California State University System, Duke University, University of Georgia, Harvard University, Howard University, Massachusetts Institute of Technology, University of Michigan, University of Mississippi, Ohio State University, University of Oxford, Institute of Political Science, Texas A&M University, as well as Boston Children’s Hospital, Boston Public Library, and OpenAI.
OpenAI Plans to Launch AI Agents at Price of $20,000
Reports indicate that OpenAI plans to charge up to $20,000 monthly for specialized AI agents.
The company is set to introduce multiple “agent” products designed for various applications, including sorting and classifying potential customers and software engineering.
Anthropic raises $3.5 BLN to Fund Its AI Ambitions
The AI startup Anthropic announced on Monday that it has raised $3.5 billion at a valuation of $61.5 billion post-funding, led by Lightspeed Venture Partners.
According to Crunchbase, this Series E funding round, which also included Bessemer Venture Partners, Cisco Investments, D1 Capital Partners, Fidelity Management & Research Company, General Catalyst, Jane Street, Menlo Ventures, and Salesforce Ventures, brings the total amount raised by the company to $18.2 billion.
The company stated in a blog post: “With this investment, Anthropic will enhance its development of next-generation AI systems, expand its computing capacity, deepen its research in interpretability and mechanical alignment, and accelerate its international expansion.” “Anthropic focuses on developing AI systems that can function as true collaborators, working alongside teams to tackle complex projects, synthesize information across domains, and help organizations achieve significant impact.”
