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.
Google Sheets Receives Upgrade Powered by Gemini for Faster Data Analysis and Visualization Creation
Google is updating its Sheets application with a new version based on Gemini, designed to help users analyze data more quickly and transform spreadsheets into charts using AI.
With this update, users can access Gemini’s capabilities to generate insights from their data, such as correlations, trends, and more. Users can now also create advanced visualizations, like heat maps, which can be inserted as static images over cells in their spreadsheets.
While the company announced the update last month, Google stated on Friday that it is now available to all Workspace commercial users.
Meta Plans to Launch Standalone AI Chat App
Reports indicate that Meta is planning to release a standalone app for its AI assistant, Meta AI, in an effort to better compete with AI-powered chat programs like OpenAI’s ChatGPT and Google’s Gemini.
According to CNBC news network, Meta may launch the standalone Meta AI app in the company’s next financial quarter (April-June). Currently, Meta AI is available to users only through a website and a suite of Meta applications, including Facebook and WhatsApp. Meta also plans to test a paid subscription service for Meta AI that will add unspecified capabilities to the assistant.
Figure to Begin Initial Testing of Its Humanoid Robot in 2025
Figure plans to bring humanoid robots to homes sooner than expected. CEO Bret Adcock confirmed on Thursday that the startup will begin “alpha testing” its Figure 02 robot at home later in 2025.
The executive states that the accelerated timeline is the result of the company’s “layman” Vision-Language-Action (VLA) model, called Helix.
Adcock’s comments come just a week after Figure announced its machine learning platform. Helix is designed to process both visual data and natural language input to accelerate the speed at which the system can learn new tasks. Earlier this month, Figure revealed that it ended its highly publicized partnership with OpenAI in favor of its own AI models like Helix.
ElevenLabs Launches Its Own Speech-to-Text Model
ElevenLabs, an AI startup that raised $180 million in a massive funding round, is known for its expertise in voice generation. The company has taken a step in a different technological direction by launching its first standalone speech-to-text model called Scribe.
The startup, valued at approximately $3.3 billion, has helped many other companies offer text-to-speech services through its vast library of voices. However, the company is now looking to enter the speech recognition field and compete with the likes of Gladia, Speechmatics, AssemblyAI, Deepgram, and OpenAI’s Whisper models, according to TechCrunch.
The Scribe model from ElevenLabs supports over 99 languages at launch. The company categorizes more than 25 languages in the model’s excellent accuracy category, where the word error rate is below 5%. This list includes English (97% accuracy), French, German, Hindi, Indonesian, Japanese, Kannada, Malayalam, Polish, Portuguese, Spanish, and Vietnamese. Other languages are classified into various categories with higher word error rates (5%-10% error rates), good (10%-20%), and moderate (25%-50%).
Alibaba Invests $52 bln in AI
Chinese holding group Alibaba announced that it plans to invest at least 380 billion yuan ($52.44 billion) in its cloud computing and AI infrastructure over the next three years, according to Reuters.
Alibaba aims to become a key partner for companies developing and applying AI in the real world as AI models evolve and their increasing need for computing power arises.
Alibaba Group, the Chinese e-commerce giant, confirmed it has achieved the fastest quarterly revenue growth rate in over a year, supported by its growing cloud computing business and AI investments.
DeepSeek Accelerates the Launch of Its New AI Model
Chinese AI startup DeepSeek is accelerating the launch of its new AI model, which will be the successor to its “R1” model released in January that made global waves.
According to sources reported by Reuters, DeepSeek was planning to launch the “R2” model in early May but now wants to release it as soon as possible.
The company hopes the new model can improve programming and be capable of thinking in languages other than English.
Competitors are still absorbing the implications of the “R1” model, which was built with less powerful chips from Nvidia but is capable of competing with those developed at a cost of hundreds of billions of dollars by American tech giants.
OpenAI Unveils Its Largest Model, “GPT-4.5”
On Thursday, OpenAI revealed its latest AI model, the long-awaited “GPT-4.5.”
GPT-4.5 is OpenAI’s largest model to date, trained using more computing power and data than any of the company’s previous releases.
Despite its size, OpenAI does not consider the “GPT-4.5” model to be a groundbreaking one; it described it as “the largest and best chat model we have yet,” according to the company’s statement.
OpenAI stated that early tests showed interactions with the “GPT-4.5” model feel more natural, noting that its broader knowledge base and improved ability to follow user objectives make it useful for tasks like writing enhancement, programming, and solving practical problems.
Amazon’s New Alexa+ Device Introduces AI-Powered “Exploration” and “Storytelling” Features for Kids
As part of the unveiling of its new AI-powered assistant Alexa+, Amazon announced it will launch two new features designed for children called “Explore with Alexa” and “Stories with Alexa.” These features, which leverage Alexa’s new AI capabilities, will be available to Amazon Kids+ subscribers.
The company showcased the new features at a press conference held in New York, emphasizing that they are designed to help children explore fun topics and encourage imaginative thinking.
The new Explore feature allows children to ask questions like, “Can plants talk to each other?” to which Alexa will respond, “Plants communicate, but not by talking.” Or they might ask a question like, “The rose is red and grows in buds. True or false?”
Anthropic Launches New AI Model With On-Demand Thinking Feature
In a new step towards AI development, Anthropic has launched its latest model, Claude 3.7 Sonnet, which features the ability to think as long as users want, making it the first hybrid reasoning model in the industry.
Unlike traditional models, Claude 3.7 Sonnet allows users to choose the level of reasoning, enabling it to provide quick answers or think longer for more accurate and in-depth responses.
The model will be available to all users and developers, but advanced reasoning features will be limited to subscribers of the paid Claude plans, while free users will receive the standard version of the model, which is more advanced than the previous Claude 3.5 Sonnet.
A New Type of AI Model Breaks the Speed Barrier in Text Generation
Inception Labs, based in California, recently unveiled a new type of large language model that operates using diffusion technology, capable of generating text faster than traditional models, named “Mercury.”
Unlike traditional models that generate text word by word, as seen in ChatGPT, diffusion-based models create responses all at once and then refine them from their initial form into a more coherent text.
Traditional large language models generate text from left to right, word by word, using a technique known as autoregression, where each word must wait for its predecessors before it can appear.
