AI-supported Cancer Care
Highlights
1. Runway introduced Gen-3, an improved video generation model.
2. Claude 3.5 Sonnet is claimed to outperform GPT-4o and Gemini in some benchmarks.
3. TikTok is enhancing its ad suite with AI-generated avatars and language dubbing tools to help brands and creators reach global audiences.
4. New methods are developed using entropy-based uncertainty estimators to detect when a prompt is likely to produce hallucination.
5. Color Health’s new AI-driven copilot accelerates cancer care by generating personalized treatment plans and identifying missing diagnostics.
Innovation Insights
1. Latent Expertise: Everyone is in R&D (One Useful Thing)
The first wave of AI has been successful in individual applications, but the second wave will require integrating AI into organizations to realize true productivity gains, a process that will take longer and needs a deeper understanding of AI’s specific strengths and weaknesses. Companies should focus on discovering AI’s potential through R&D and expert use rather than viewing it solely as a cost-cutting tool. Experts play a crucial role in unlocking AI’s latent expertise, and sharing their insights will be essential for maximizing AI’s benefits and mitigating its risks.
2. Six steps for launching predictive AI projects (Ideas Made to Matter)
In “The AI Playbook,” Eric Siegel outlines how organizations can bridge the gap between data scientists and business stakeholders to successfully deploy predictive machine learning projects, advocating for a “BizML” approach that includes six key steps: establishing deployment and prediction goals, setting the right metrics, preparing the data, training the model, and deploying it effectively. By fostering collaboration and aligning technological efforts with business objectives, companies can better leverage AI for impactful outcomes.
3. The Pragmatist’s Guide to GenAI in E-Commerce (BCG)
E-commerce executives are advised to focus on practical build-or-buy technology decisions to leverage the rapid evolution of generative AI, considering the role of external providers and custom solutions. The decision to build in-house should be based on the need for a differentiated customer experience using proprietary data. The implications of these build decisions involve understanding their impact on operational processes, data usage, and team roles.
4. AI Survey: Four Themes Emerging (Bain)
In 2024, many companies are still building their generative AI solutions instead of relying on off-the-shelf ones. Many executives are focused on delivering real value, passing the stage of being concerned with quality and capabilities.AI use cases seem more successful in sales, software development, marketing, customer service, and customer onboarding than in legal, operations, and HR.
AI Innovations
1. AI models
Runway’s new AI, Gen-3, improves upon previous models with faster, higher-quality video generation from text and images, offering enhanced control over video structure, style, and motion for content creators (TechCrunch).
Apparate Labs has released PROTEUS, a real-time model for generating highly realistic and expressive human avatars, designed to enhance interactive experiences by enabling instantaneous, visually expressive, and voice-controllable artificial conversational entities (Apparate AI).
Meta’s FAIR team released several AI models, including Chameleon for mixed-modal understanding, multi-token prediction for efficient LLM training, JASCO for controlled music generation, AudioSeal for AI speech detection, and tools to increase diversity in image generation, to foster responsible innovation and collaboration in AI (Meta).
Anthropic has launched Claude 3.5 Sonnet, its fastest and most intelligent AI model yet, which outperforms competitors like OpenAI’s GPT-4o and Google’s Gemini in various benchmarks, and includes a new feature called Artifacts for interactive results editing directly within the app (The Verge).
2. TikTok
TikTok is enhancing its ad suite with AI-generated avatars and language dubbing tools to help brands and creators reach global audiences by providing customizable digital avatars and multi-language support (The Verge). It has also introduced Symphony, an AI assistant integrated with Adobe Express and TikTok Ads Manager, designed to streamline and enhance content creation by automating tasks like script generation, idea development, video creation, and optimization (Yahoo).
3. NVIDIA
NVIDIA introduced Omniverse Cloud Sensor RTX™, a suite of microservices that use realistic sensor simulation in virtual environments to accelerate the development and testing of autonomous vehicles, robots, and smart spaces, enhancing safety and efficiency (NVIDIA).
4. Cancer Care
Color Health’s new AI-driven copilot, powered by GPT-4o, accelerates cancer care by generating personalized treatment plans and identifying missing diagnostics, enhancing clinicians’ ability to make timely, evidence-based decisions (OpenAI).
5. Detecting hallucinations
New methods are developed using entropy-based uncertainty estimators to detect when a prompt is likely to produce a confabulation, i.e. arbitrary and incorrect generation (Nature).
6. Gaming
Generative AI could revolutionize gaming by enabling NPCs to engage in unscripted, dynamic interactions, creating immersive and ever-evolving game experiences that redefine traditional gameplay (MIT Tech Review).
7. Video to audio
Runway’s new video-to-audio (V2A) technology uses video pixels and text prompts to generate synchronized soundtracks for silent videos, providing enhanced control over audio output and enabling creative soundscapes for various types of footage (Google).
8. Ilya Sutskever
The former chief scientist and co-founder of OpenAI, has launched Safe Superintelligence Inc. (SSI) to focus on advancing AI capabilities while prioritizing safety (TechCrunch).
9. AI voice cloning
Universal Music Group is partnering with AI startup SoundLabs to offer its artists AI voice cloning technology, allowing them to create, own, and use their voice models for various applications, while emphasizing ethical use and protecting artists’ virtual likenesses (Rolling Stone).
Other Innovations
1. Supersonic flight
Boom Supersonic’s North Carolina superfactory will begin producing 33 Overture supersonic jets annually by the end of 2024, doubling to 66 jets later, promising to revolutionize air travel with New York to London flights in 3.5 hours (SupercarBlondie).
2. Cancer treatment
The ratio between bacterial communities in the gut might predict a positive response to cancer drugs called checkpoint inhibitors (Nature).
3. Solid-state batteries
TDK claims a major breakthrough in solid-state battery technology with new ceramic materials that offer 100 times the energy density of its current batteries, potentially revolutionizing small consumer electronics by providing longer operation times and enabling smaller device sizes (Ars Technica).
Chinese automakers, including Nio and WeLion, are advancing the development of semi-solid-state batteries with gel electrolytes, offering improved energy density and shorter-term availability compared to full solid-state batteries, which face more significant production challenges (IEEE Spectrum).
4. CarPlay
Apple’s next-generation CarPlay, designed to wirelessly integrate with car systems and unify interfaces across vehicle screens, allows automakers to customize its features extensively (The Verge).
5. RISC-V laptop
The Framework Laptop 13 is set to become one of the first laptops with a RISC-V processor, offering an open-source alternative to Intel and AMD, though it’s currently best suited for developers and hobbyists due to its lower performance and limited features (The Verge).
6. Internet service
SpaceX has introduced the “Starlink Mini,” a backpack-sized satellite internet antenna for $599, which offers portable, high-speed internet service at $150 per month for 50 GB of data, aiming to make connectivity more accessible and affordable globally (CNBC).
