• AI KATANA
  • Posts
  • AI humanoid robots inch their way toward the workforce

AI humanoid robots inch their way toward the workforce

Also: Meta’s Llama Incubator Program Accelerates AI Adoption for Startups and SMEs in Singapore

Author

AI KATANA
February 23, 2025

Hello!

Today’s newsletter highlights significant strides in AI technology and its growing impact across various industries. Meanwhile, Meta’s collaborative Llama Incubator Program in Singapore aims to fast-track AI adoption among startups and SMEs, equipping them with tools and mentorship to effectively implement AI solutions. In surprising findings, Israeli researchers identified patterns of cognitive decline in prominent AI models, mirroring human neurodegeneration, raising questions about their reliability in critical applications. Additionally, IEEE is addressing the educational gap in AI with specialized training courses and a Mini MBA program designed to empower professionals to strategically integrate AI in business contexts. Chinese universities are also embracing the AI boom, launching specialized curricula centered around DeepSeek’s advanced models, reflecting China’s commitment to leading global education standards by 2035. In AI tool developments, Google has expanded availability of its Deep Research tool to Gemini Advanced users, significantly streamlining complex topic exploration tasks. Finally, Together AI has reached a substantial valuation of $3.3 billion following a successful funding round, underscoring the global momentum towards accessible, powerful open-source AI infrastructures.

Sliced just for you:

  • 🤖 AI humanoid robots inch their way toward the workforce

  • 🚀 Meta’s Llama Incubator Program Accelerates AI Adoption for Startups and SMEs in Singapore

  • 🧠 Scientists Tested AI For Cognitive Decline. The Results Were a Shock.

  • 📚 IEEE Offers AI Training Courses and a Mini MBA Program

  • 🎓 Chinese universities launch DeepSeek courses to capitalise on AI boom

AI-powered humanoid robots are progressively entering the workforce, driven by advancements in generative AI and mechatronics, promising significant transformations across various industries. Prominent tech vendors such as Tesla’s xAI, Boston Dynamics, Agility Robotics, and Figure.ai are spearheading developments aimed at making humanoid robots commercially viable within two years. Companies like logistics firm GXO have already deployed Agility Robotics’ Digit robots in practical warehouse scenarios and are conducting tests with other robot brands for tasks like picking, packing, and palletizing. Elon Musk’s xAI has introduced the Optimus humanoid, which leverages the powerful Grok-3 AI model and is anticipated to be operational soon. While these innovations suggest humanoid robots could comprise about 10% of smart robots sold by 2027, significant hurdles remain, including high production costs, technological challenges in achieving human-like dexterity and sensory capabilities, and unresolved AI safety concerns. Industry experts recognize potential applications in hazardous environments, manufacturing, logistics, healthcare, retail, and construction, though widespread adoption hinges on addressing these practical limitations. Despite skepticism around timelines and feasibility, investments and collaborations continue, signaling an enthusiastic push toward integrating humanoid robots as key players in future workplace automation.

Singapore has launched the Llama Incubator Program, an initiative driven by Meta in collaboration with prominent local organizations including Enterprise Singapore, IMDA, GovTech, SGInnovate, AI Singapore, Deloitte, and coordinated by e27. The program aims to accelerate AI adoption among 40 digitally prepared startups and SMEs, leveraging Meta’s open-source Llama large language model to enhance products, workflows, and market opportunities. Structured into two tracks—one tailored for startups developing AI-driven Minimum Viable Products (MVPs), and another for SMEs integrating AI into their operations—the incubator provides technical resources, mentorship, and strategic guidance. Eligibility for startups includes a minimum operational period of one year, secured funding of at least SGD 250,000, and demonstrable technological proficiency, while SMEs should have less than SGD 100 million revenue, fewer than 200 employees, and a clear, leadership-backed AI use case. The incubation spans from April to October 2025, culminating in a public Demo Day and a showcase at the SWITCH event. Selection criteria emphasize digital readiness, the practicality of the business case, innovation potential, and responsible AI practices. This collaborative effort represents a significant opportunity for regional businesses to innovate and drive sustainable growth through AI technology.

🧠 Scientists Tested AI For Cognitive Decline. The Results Were a Shock.

Recent research from Israel reveals an unexpected similarity between human cognitive decline and the performance deterioration observed in prominent AI models, including OpenAI’s ChatGPT, Alphabet’s Gemini, and Anthropic’s Claude. Using cognitive assessments traditionally employed by neurologists, such as the Montreal Cognitive Assessment (MoCA), the researchers discovered that these large language models (LLMs) displayed patterns of cognitive impairment that mirror human neurodegenerative processes. ChatGPT 4o achieved the highest score with a mild impairment rating, while Alphabet’s Gemini scored alarmingly low, indicating severe cognitive deficits. Across all models tested, significant weaknesses emerged particularly in visuospatial and executive function tasks, such as copying geometric shapes and performing spatial orientation challenges, along with empathy-related tests reminiscent of frontotemporal dementia in humans. Although these findings emphasize limitations and caution against relying too heavily on current AI for critical applications like clinical diagnostics, they also underscore the potential for future AI advancements to mitigate these cognitive shortcomings.

IEEE is addressing the growing significance of AI in the business world by offering specialized educational programs designed to help businesses strategically implement AI technologies for enhanced efficiency, growth, and competitive advantage. Acknowledging that despite increasing integration, AI remains poorly understood among employees, IEEE has developed comprehensive training courses aimed at equipping professionals with vital practical skills. These include specialized course series such as “AI and Machine Learning in Chip Design,” focusing on design automation and future strategies, and “Integrating Edge AI and Advanced Nanotechnology in Semiconductor Applications,” covering the convergence of edge computing and nanotech. Another offering, “AI Integration in Semiconductor Manufacturing,” highlights how AI improves production efficiency and product quality. Each course grants participants professional development credits and digital credentials suitable for social media sharing. Furthermore, IEEE has collaborated with Rutgers University to launch an online Mini-MBA in AI, tailored to enable employees to strategically apply AI across industry-specific applications, optimizing operations, and enhancing organizational success.

Chinese universities are rapidly integrating AI education into their curricula by launching specialized courses based on the advanced models from the Chinese AI startup, DeepSeek. This initiative aligns with China’s strategic focus on technological innovation within educational institutions to foster economic growth. DeepSeek, based in Hangzhou, has gained international acclaim for its models, DeepSeek-V3 and DeepSeek-R1, considered competitive with leading global AI models from OpenAI and Meta. Notably, Shenzhen University has initiated courses that explore DeepSeek’s technologies, including important discussions on ethics, security, and privacy implications, emphasizing the balance between innovation and ethical norms. Other leading institutions, such as Zhejiang University, Shanghai’s Jiao Tong University, and Renmin University of China, have also adopted DeepSeek to enhance AI education, support research, and improve administrative operations. These developments come in response to China’s first national education action plan aiming to position its education system among the world’s best by 2035. This drive is further highlighted by DeepSeek founder Liang Wenfeng’s recent high-profile engagement with President Xi Jinping and other Chinese tech industry leaders, signaling strong governmental endorsement of AI development.

🛠️ AI tools updates

Google has expanded access to its innovative AI tool, Deep Research, making it available to Gemini Advanced users via Google Workspace. Deep Research accelerates complex topic exploration by rapidly analyzing extensive internet data, synthesizing thousands of sources into comprehensive and easily digestible reports within minutes—tasks that would typically take human researchers hours to complete. Its versatility is showcased across various sectors, proving invaluable for conducting industry-specific research, competitive analysis, and tailored customer insights. Educational professionals can also leverage Deep Research for lesson planning, developing classroom projects, presentations, and grant applications. The tool functions by initially creating a customizable, multistep research plan and dynamically refining its analysis as it uncovers new relevant information. Users receive a detailed report with key findings, along with direct links to original sources, and can further delve into topics with follow-up queries. Currently accessible only through web interfaces for Gemini Advanced users, Google plans to introduce mobile support and extend language capabilities beyond English soon. Additionally, alongside Deep Research, Google unveiled new experimental Gemini models such as 2.0 Flash Thinking Experimental and 2.0 Pro Experimental, recognized as some of the most powerful generative AI models globally.

💵 Venture Capital updates

Together AI, an AI startup based in San Francisco, has achieved a remarkable valuation of $3.3 billion following its latest funding round, raising $305 million led by General Catalyst and Prosperity7, the venture capital fund of Aramco Ventures, with significant contributions from Salesforce Ventures, Nvidia, and Coatue. Established in 2022, Together AI provides a cloud platform enabling developers and researchers to efficiently train and deploy generative AI models. The company has rapidly gained popularity, partly due to its strategic provision of highly-demanded Nvidia GPUs. Supporting over 200 open-source AI models, including Meta’s Llama and China’s DeepSeek, Together AI has expanded its developer user base to more than 450,000. The fresh capital will primarily fuel the expansion of Together AI’s cloud infrastructure and large-scale deployment of Nvidia’s cutting-edge Blackwell graphics processing units. The funding underscores Together AI’s role in promoting secure, accessible, and powerful open-source AI technologies at a global scale.

🫡 Meme of the day

⭐️ Generative AI image of the day