Brain organoids-based processor
Highlights
1. Recent advancements in AI, particularly multimodal models, are significantly enhancing robotic capabilities.
2. AI can enhance leadership by promoting awareness, wisdom, and compassion.
3. Stable Audio Open is a new text-to-audio model that can create 47 seconds of audio.
4. KLING from Kuaishou will likely compete with Sora.
5. Qwen 2-72B from Alibaba is a top open-source LLM.
6. Scientists used AI to analyze global microbial diversity, predicting nearly 1 million potential new antibiotics.
7. FinalSpark developed a bioprocessor that is part of an online platform with remote access to 16 human brain organoids.
Innovation Insights
1. Robots are suddenly getting cleverer. What’s changed? (The Economist)
Recent advancements in AI, particularly multimodal models, are significantly enhancing robotic capabilities, allowing robots to understand and manipulate their environments through a combination of text, images, and sensor data. These improvements have led to more intuitive and adaptable robots that can perform complex tasks, explain their actions, and be reprogrammed with simple text prompts. This progress has sparked renewed interest and investment in robotics, marking a shift toward “embodied AI” where software advancements are increasingly applied to physical machines.
2. How AI can make us better leaders (Harvard Business Review)
Humans excel at creating transformative tools like AI but often struggle with the resulting changes. AI, despite its potential risks, can enhance leadership by promoting core human qualities—awareness, wisdom, and compassion—enabling leaders to blend technological power with humanity. To harness AI effectively, leaders should adopt a “both/and” mindset, integrating AI’s capabilities while focusing equally on developing their intrinsic human strengths.
AI Innovations
1. AI tools
ChatGPT Edu is a specialized, affordable version of ChatGPT built for universities to responsibly deploy AI across campus with advanced capabilities, enterprise-level security, and extensive language support (OpenAI).
ElevenLabs’ new Sound Effects AI tool allows users to generate specific sound effects for various applications like podcasts, movies, or games by simply writing a prompt, offering up to 22 seconds of customizable audio clips (The Verge).
A new AI-powered monitoring system, resembling an airplane black box, is being developed to improve surgical safety by recording and analyzing operating room activities. While this technology shows promise in reducing errors and enhancing procedural efficiency, it also raises privacy and legal concerns, leading some surgeons to resist or sabotage its implementation (MIT Technology Review).
Meta has introduced a new AI assistant for businesses on WhatsApp to enhance customer interactions and boost sales for businesses (Bloomberg).
Jasper AI has unveiled Flash Diffusion, a novel approach that speeds up diffusion models, enabling quicker and higher-quality image generation (Link).
Asana has introduced AI teammates designed to enhance workflow management and teamwork, offering customizable AI solutions that advise on priorities, execute tasks, and adapt to specific processes (Asana).
Wix has introduced a generative AI feature in its app builder that allows users to create and edit iOS and Android apps by describing their desired app in plain English (TechCrunch).
2. New models
Microsoft launches AI weather prediction model ‘Aurora,’ the first extensive foundation model for atmospheric analysis (AIM).
Stable Audio Open is an open source text-to-audio model that generates up to 47 seconds of audio samples, such as drum beats and sound effects, enabling fine-tuning on custom data for sound designers and musicians (Stability AI).
Chinese tech company Kuaishou has launched KLING, an advanced AI model that can create high-quality videos up to 2 minutes in length, potentially competing with OpenAI’s forthcoming Sora (Kuaishou).
Alibaba’s open-source Qwen 2-72B model has secured the top position on the Hugging Face Open LLM Leaderboard (Link).
3. AI for science
Scientists at the University of South Florida are using AI and virtual reality to create detailed 3D brain maps of mice to advance understanding and treatment of Alzheimer’s, autism, and brain injuries (The Guardian).
Scientists used AI to analyze global microbial diversity, predicting nearly 1 million potential new antibiotics and accelerating the discovery process significantly. This research, published in the journal Cell, marks a major advance in combating antibiotic resistance (The Guardian).
4. Nvidia
NVIDIA has announced new RTX technology and tools, including Project G-Assist and ACE NIMs, to enhance AI assistants, digital humans, and generative AI experiences on GeForce RTX AI laptops and Windows PCs (Nvidia). NVIDIA has made its ACE digital human microservices, including tools for speech recognition, facial animation, and new gesture generation, generally available for cloud and early access for RTX AI PCs, enabling lifelike digital avatars in customer service, gaming, and healthcare applications (Nvidia).NVIDIA’s upcoming Rubin architecture will succeed the Blackwell family of GPUs and CPUs in 2026, featuring HBM 4 memory and a 4x reticle design (ZDNet).
5. Extracting concepts from GPT-4
Researchers developed scalable methods to break down GPT-4’s internal workings into 16 million often interpretable patterns, offering insights into its neural activity and sharing our findings, code, and visualizations with the research community to promote further understanding (OpenAI).
Other Innovations
1. Brain organoids-based processor
FinalSpark’s bioprocessor is part of an online platform with remote access to 16 human brain organoids. It is claimed to consume a million times less power than traditional processors and can learn and process information, though scalability and shared processing viability remain uncertain (MSN).
2. GPU alternative
A prototype circuit using variable resistors has been developed by researchers at the University of Pennsylvania to offer a more energy-efficient alternative to GPUs for machine learning. By encoding data as voltages instead of binary values, this circuit performs tasks like data classification with lower energy consumption (MIT Technology Review).
3. Solid refrigerants
Pascal has developed solid refrigerants that enhance HVAC efficiency by 50-80% without contributing to global warming, using existing infrastructure (Springwise).
4. Night-vision lenses
A breakthrough by TMOS researchers in Australia has enabled the creation of ultra-thin, lightweight night-vision lenses, potentially transforming nighttime activities by making it possible to see in the dark through ordinary glasses (New Atlas).
5. Starship
SpaceX’s Starship successfully launched and performed its first controlled splashdown (Ars Technica).
6. Smart contact lenses
Scientists have developed an ultra-thin battery for smart contact lenses that can be powered by tears (CNBC).
