👋If you are a new reader, my name is Danar Mustafa. I write about product management focusing on AI, tech, business and agile management. You can visit my website here or visit my Linkedin here.
My goal is to be your guide through the exciting world of technology, startups, and our digital society. I’m here as a tech enthusiast to explore the latest advancements that are shaping our world.
📌 February Infographic:Comparing Humanoid Robots Across The World📌
OpenAI and SoftBank have partnered to create ‘Cristal Intelligence,’ investing $3 billion annually to develop AI solutions across SoftBank’s portfolio, aiming to revolutionize enterprise AI applications in Japan. The partnership includes the formation of SB OpenAI Japan, a joint venture focused on marketing and implementing AI solutions tailored for Japanese enterprises, emphasizing data security with Japan-based servers.
Cristal Intelligence aims to automate over 100 million workflows within SoftBank’s companies, enhancing productivity and operational efficiency.
https://opentools.ai/news/openai-and-softbank-join-forces-to-bring-cristal-intelligence-to-japan
🧠 EU invests $56M to build open-source AI rival 🧠
The European Commission has awarded a pioneering AI project aimed at developing large language models specifically tailored for European languages. This initiative is part of a broader strategy to enhance digital capabilities across Europe.
The main goal of the project is to create open-access large language models that can support multiple European languages. This will facilitate better communication and understanding across diverse linguistic backgrounds, promoting inclusivity in digital spaces.
This project is significant as it addresses the challenges faced by non-English speakers in accessing information and services online. By focusing on European languages, the initiative aims to ensure that all citizens can benefit from advancements in AI technology.
The implementation will involve collaboration among various stakeholders, including academic institutions, industry leaders, and governmental bodies. The project emphasizes transparency and accessibility, ensuring that the developed models are available for public use and further research.
The expected outcomes include improved natural language processing capabilities for European languages, fostering innovation in AI applications across different sectors. Additionally, this initiative is anticipated to contribute to the preservation and promotion of linguistic diversity within Europe.
🧠 The Beatles won a Grammy with an AI-enhanced song 🧠
The Beatles have achieved a historic milestone by winning their eighth Grammy award for the song “Now and Then,” which is notable for being the first known instance of artificial intelligence (AI) being used in the creation of a Grammy-nominated song. The track won the award for Best Rock Performance, competing against prominent bands like Green Day, Pearl Jam, The Black Keys, Idles, and St. Vincent. https://newsgpt.ai/2025/02/04/the-beatles-make-history-with-ai-enhanced-grammy-winning-song-now-and-then/
🧠 Microsoft is forming a new unit called the Advanced Planning Unit (APU) 🧠
Microsoft is establishing an “Advanced Planning Unit” aimed at studying the implications and future scenarios of artificial intelligence (AI). The new unit will operate under the Office of Microsoft AI’s CEO and will focus on several key areas:
1. Conducting cutting-edge research both within Microsoft and externally.
2. Exploring various potential future scenarios regarding AI development and its impact on society.
3. Crafting concrete product recommendations based on their findings.
4. Suggesting planning outcomes that align with their research insights.
https://www.theverge.com/news/604673/microsoft-is-forming-an-advanced-planning-unit-to-study-ai
🧠 OpenAI’s GPT-4.5 Released 🧠
OpenAI has introduced its latest AI model, GPT-4.5, which is described as the “best model yet” for chat purposes.
GPT-4.5 includes advancements in unsupervised learning that enhance its ability to recognize patterns, draw connections, and generate creative insights. It provides a more natural conversational experience with an improved personality.
The new model outperforms its predecessor, GPT-4o, across various categories such as everyday queries, professional inquiries, and creative tasks. It boasts a broader knowledge base and a better understanding of user intent.
https://openai.com/index/introducing-gpt-4-5/
🧠 Meta introduced Aria Gen 2 glasses 🧠
Meta has unveiled its second-generation smart glasses, the Aria Gen 2, which are designed to enhance augmented reality experiences. These glasses come with several new features and improvements over their predecessor.
The Aria Gen 2 glasses are equipped with advanced sensors that improve their functionality. This includes enhanced cameras for better image capture and improved environmental awareness. The integration of these sensors allows for more immersive augmented reality applications, making the glasses suitable for a variety of uses, from social interactions to professional applications.
🧠 Amazon unveiled Alexa+ 🧠
Amazon has introduced Alexa+, a next-generation assistant enhanced by generative AI, designed to provide a more conversational and personalized experience for users. With over 600 million Alexa devices in use, the advancements aim to improve customer interactions and make daily tasks easier.
Conversational Abilities Alexa+ offers expansive and natural conversations, understanding colloquial expressions and complex ideas. This makes interactions feel more like engaging with a friend rather than technology.
https://www.aboutamazon.com/news/devices/new-alexa-generative-artificial-intelligence
🧠 Claude 3.7 Sonnet with ‘hybrid reasoning’ 🧠
Claude 3.7 Sonnet is the latest hybrid reasoning model developed by Anthropic, recognized as their most intelligent model to date. It incorporates advanced capabilities in coding, content generation, data analysis, and planning.
The release of Claude 3.7 Sonnet occurred on February 24, 2025, following the previous version, Claude 3.5 Sonnet. The new model offers significant improvements over its predecessor in various applications.
Claude 3.7 Sonnet is accessible through the Anthropic API, Amazon Bedrock, and Google Cloud’s Vertex AI for developers looking to create custom AI solutions. For general users, it can be accessed via Claude.ai on web and mobile platforms. Pricing starts at $3 per million input tokens and $15 per million output tokens, with potential savings through prompt caching and batch processing.
https://www.anthropic.com/claude/sonnet
🧠1X’s NEO Gamma home humanoid 🧠
NEO Gamma, developed by 1X Technologies, is a cutting-edge personal assistant robot designed to transform home environments. It can perform various tasks such as tidying up, deep cleaning, and managing household functions. The robot’s hands are engineered for safe and precise interactions, making it suitable for delicate environments.
Beyond its functional capabilities, NEO Gamma serves as a companion. It can engage in conversations, assist with homework, and collaborate on tasks. Its soft-knit suit allows for flexible movements and comfortable interactions, enhancing its role as both a helper and a partner in the household.
🧠 Figure debuts new system for household robots 🧠
Helix is a groundbreaking Vision-Language-Action (VLA) model developed to enhance robotics by integrating perception, language understanding, and control. It represents several firsts in the field, including full upper-body control of humanoid robots, multi-robot collaboration capabilities, and the ability to pick up various household objects based on natural language prompts.
Training involves a diverse dataset collected from teleoperated behaviors with natural language instructions generated through an auto-labeling VLM. The model is trained end-to-end without requiring task-specific adaptations.
The inference process is designed for efficiency on low-power embedded GPUs, allowing real-time operation by splitting tasks between S2 (high-level planning) and S1 (low-level control).
🧠 Google’s multi-agent AI co-scientist 🧠
Google has introduced an innovative AI system called the AI co-scientist, which is built on the Gemini 2.0 platform. This tool is designed to assist scientists in formulating new hypotheses and developing research plans.
The AI co-scientist allows researchers to articulate their research goals using natural language. For instance, a scientist might want to investigate the spread of a disease-causing microbe. In response, the AI will generate testable hypotheses, provide a summary of relevant published literature, and suggest possible experimental approaches.
It is important to note that the AI co-scientist serves as a collaborative tool aimed at enhancing the scientific process rather than automating it. The goal is to support experts in gathering information and refining their research efforts.
https://blog.google/feed/google-research-ai-co-scientist/
🧠 Global tensions rise at Paris AI summit 🧠
At the Grand Palais, leaders from nearly 100 countries gathered for the Paris Artificial Intelligence Action Summit, co-hosted by French President Emmanuel Macron and Indian Prime Minister Narendra Modi. The summit aimed to shape future AI governance, but differing regulatory approaches and national interests were evident.
1. U.S. and U.K. Refusal to Sign AI Declaration The U.S. and U.K. did not sign a declaration supported by 61 nations that committed to “inclusive and sustainable” AI. U.S. Vice President J.D. Vance criticized the notion of inclusive AI, reflecting a broader trend of scaling back diversity initiatives in tech companies. The U.K.’s refusal was attributed to concerns over national security rather than alignment with the U.S., despite its previous commitment to AI safety.
2. China’s Willingness to Share AI Achievements Chinese Vice Premier Zhang Guoqing expressed China’s readiness to collaborate internationally on AI development, emphasizing a diplomatic approach that aligns with President Xi Jinping’s vision for global cooperation in technology.
3. Europe’s Commitment to Significant AI Investments Europe announced substantial financial investments in AI, including the launch of InvestAI, which aims to mobilize €200 billion for development, with specific funds allocated for training large models. Additionally, over 60 European companies initiated the EU AI Champions Initiative to accelerate adoption and innovation while simplifying regulations.
4. Launch of Online Safety Tools A coalition including Google, OpenAI, Roblox, and Discord introduced the Robust Open Online Safety Tools (ROOST) initiative at the summit. This initiative aims to provide free open-source tools for detecting and reporting child sexual abuse material online, having raised over $27 million in funding.
https://www.csis.org/analysis/frances-ai-action-summit
🧠 3 New Unicorn Startups in 2025 🧠
Netradyne is a computer vision startup founded in 2015 that specializes in developing advanced technology for fleet safety and driver monitoring. The company raised a $90 million Series D funding round, which valued it at $1.35 billion. Netradyne’s solutions leverage artificial intelligence and machine learning to analyze driving behavior and improve road safety for commercial fleets. The funding round was led by Point72 Ventures.
Hippocratic AI
Founded in 2023, Hippocratic AI is focused on creating healthcare models using artificial intelligence to enhance patient care and outcomes. The startup raised a $141 million Series B funding round, bringing its valuation to $1.64 billion. Hippocratic AI aims to develop innovative solutions that can assist healthcare providers in making data-driven decisions while improving overall efficiency within the healthcare system. This funding round was led by Kleiner Perkins.
https://www.hippocraticai.com/
Truveta
Truveta is a genetic research company established in 2020 that utilizes real-world data to advance medical research and improve patient outcomes. The company raised a significant $320 million funding round, which valued it at $1 billion. Truveta’s mission is to harness the power of health data to provide insights into diseases and treatments while ensuring patient privacy and security. Its investors include corporate venture capital arms from Microsoft and Regeneron Pharmaceuticals.
🧠 Swedish Startup Sana 🧠
Sana, a Stockholm-based AI company, has raised over 580 million kronor in a financing round. As part of this funding, Sana has acquired the British company Ctrl, which specializes in agent technology. This acquisition is aimed at facilitating Sana’s growth in both the United States and Europe.
Sana plans to expand its workforce from 150 to 300 employees across its offices in Stockholm, London, and New York. The investment round was led by New Enterprise Associates, with participation from investors such as Menlo Ventures, Katarina Martinsson, and Sven Hagströmer.
Sana is developing a more advanced version of AI assistants capable of performing tasks across various applications and services. The founder, Joel Hellermark, noted that navigating global knowledge is easier than managing data within companies due to its often scattered and unstructured nature.
Hellermark claims that Sana currently faces no direct competition but acknowledges that many companies using Microsoft’s OpenAI-based Copilot are dissatisfied with their experiences.
Sana aims to reach one billion users globally; it currently has around one million users, including notable companies like Amazon, Robinhood, and Bristol Myers Squibb. Founded in 2016, Sana’s valuation has reached 5.3 billion kronor following the latest funding round.
🧠 Research Paper of the Week – OpenAI SWE-Lancer 🧠
OpenAI has introduced a new benchmark called SWE-Lancer, designed to evaluate AI’s coding abilities in the context of freelance software development. This benchmark assesses AI performance using over 1,400 real-world tasks sourced from Upwork, with a total payout value of $1 million.
Real Freelance Jobs, Real Challenges Unlike traditional AI coding benchmarks that focus on synthetic problems, SWE-Lancer utilizes actual freelance tasks ranging from simple bug fixes worth $50 to complex feature implementations valued at $32,000. The benchmark evaluates not only code generation but also decision-making skills akin to those of an engineering manager.
How Well Do AI Models Perform? The results indicate that even advanced AI models face significant challenges when tackling these tasks. While they can generate code snippets and assist with debugging, they struggle with the complexities of freelance engineering work that require creativity and long-term planning. This suggests that fully autonomous coding by AI is still far from realization.
To promote further research, OpenAI has made essential resources related to SWE-Lancer publicly available. This includes a unified Docker image and a subset known as SWE-Lancer Diamond for evaluation purposes. The benchmark aims to provide insights into the economic impact of AI in software engineering and contract-based work.
🧠 Use Case of the Month – Built and deploy application without coding skills 🧠
The benchmark evaluates the performance of various AI coding tools in building and deploying applications without requiring coding skills. The tools tested include Cursor, Replit, Windsurf Editor by Codeium, Cline, and Claude Code. The benchmark utilized different modes of each tool and provided a single prompt for both tasks without further prompting to maintain objectivity.
In conclusion, while these AI coding tools show potential in automating app development tasks, they currently face significant challenges in creating fully functional APIs and applications according to specified requirements.
https://research.aimultiple.com/ai-code-editor/#app-building-benchmark-results
🧠 Tools of The Month 🧠
GPT 4.5 – latest model – https://openai.com/index/introducing-gpt-4-5/
Ideogram – image generation – https://ideogram.ai/
Pika 2.2 – Video AI generation with transition mode – https://pikartai.com/pika-2-2/
🧠 AI / Tech Events in February 🧠
Mar 3-6 MWC Barcelona the world’s largest mobile technology event, featuring advancements in 5G, IoT, and connectivity – https://www.mwcbarcelona.com/
Mar 12 – 13 World Big Data & AI World, covers data and AI advancements, featuring talks on data analytics, machine learning, and business applications – https://www.bigdataworld.com/
Discover more from The Tech Society
Subscribe to get the latest posts sent to your email.