AI in Marketing

We start with OpenAI asking contractors for real work artifacts to benchmark agents against human performance, then widen out to Excel’s stubborn grip on business data and the operational risks that come with it. From there, a Wharton report and the Remote Labor Index show enterprises treating AI as an ROI question and talent problem, not a toy, even as current agents handle only a sliver of real work. We next look at transparency and trust, from X open-sourcing its recommendation code to Anthropic’s Claude Code deployment at Allianz. The focus then tilts toward AI at the edge of daily life and the physical world: AI-enhanced collectibles at CES, AI used in a mountain search-and-rescue, fusion progress on China’s “artificial sun,” and a robotics startup compressing development cycles with Nvidia’s Isaac tools. We close back in health care, where OpenAI and Anthropic are pushing deeper into medical data and workflows, raising the stakes on how safely these systems are built and governed.

OpenAI is requesting contractors to upload past work projects to assess the performance of its AI models, according to records obtained by WIRED. This initiative is part of OpenAI's strategy to establish a human baseline for tasks, allowing comparisons with AI models. The company launched a new evaluation process in September to measure AI performance against human professionals across various industries, aiming to achieve Artificial General Intelligence (AGI). Contractors are asked to provide real examples of work, such as Word documents, PDFs, or Excel files, while removing confidential and personally identifiable information. OpenAI emphasizes that the examples should reflect actual on-the-job tasks. A confidential document from OpenAI outlines a task example involving a luxury concierge company, where a contractor would upload a real itinerary created for a client. OpenAI provides a tool called "Superstar Scrubbing" to help contractors remove confidential information. However, Evan Brown, an intellectual property lawyer, warns that AI labs could face trade secret misappropriation claims if contractors inadvertently share confidential information. OpenAI, along with other AI labs like Anthropic and Google, is hiring contractors to generate high-quality training data for developing AI agents capable of automating enterprise work. The demand for skilled talent has created a lucrative sub-industry within the AI training world, with companies like Handshake and Surge reaching high valuations.

From Artificial General we shift focus to Acuity Training

Excel, despite being 40 years old, remains a staple in the business world, with two-thirds of office workers using it hourly, according to Acuity Training. While some tech leaders criticize it as a hindrance to digital workflows and AI integration, Excel's simplicity for quick data analysis and chart creation keeps it popular. However, its use can lead to poorly documented processes and data management issues, as seen in Health New Zealand's financial management and the UK's anesthetist recruitment chaos. Organizations like Telus are moving away from Excel to improve data integration and automation, but resistance remains due to users' attachment to existing setups. Alternatives like LinkSpace and online accounting packages offer more efficient data management and cost savings.

Acuity Training to Remote Labor

A new report from Wharton and GBK Collective highlights the mainstream adoption of AI in enterprises, emphasizing a shift from technology challenges to talent management. The report, based on surveys of over 800 enterprise leaders, indicates that AI is now deeply integrated into business operations, with 72% of companies measuring the return on investment (ROI) from AI initiatives. This focus on ROI and accountability marks a departure from earlier speculative adoption driven by excitement and fear of missing out. The report reveals that 60% of enterprises have appointed a Chief AI Officer or equivalent, underscoring the strategic importance of AI. However, a significant concern is the potential decline in employee skills as reliance on AI grows, with 43% of leaders worried about skill atrophy. The report suggests that companies need to focus on training pipelines that maintain core skills alongside AI usage. Additionally, the Remote Labor Index by Scale AI and CAIS shows that AI agents struggle with complex tasks, completing only 2-3% of freelance jobs. This underscores the gap between AI's current capabilities and the vision of a fully autonomous AI workforce, highlighting the need for continued human oversight and skill development.

Moving from Remote Labor to the world of Elon Musk

Elon Musk's social media platform X, formerly known as Twitter, has announced the open-sourcing of its code for the "For You" timeline recommendation algorithm. This move aims to enhance transparency by allowing the public to view, use, and modify the code, excluding parts that could compromise user safety or platform security. The open-sourcing initiative aligns with a broader industry trend in 2025, where tech companies face increasing pressure to disclose how algorithms shape online experiences. By releasing the code, X invites external contributions and scrutiny, potentially accelerating improvements in personalization and content moderation. This decision positions X alongside major tech companies like Google and Microsoft, which maintain open-source projects, and reflects a shift towards algorithmic transparency. While sensitive code related to user safety and ad targeting remains private, the release of the core recommendation system could reshape user expectations and set a precedent for social platforms sharing their inner workings. Open sourcing benefits creators through community contributions and users by providing free access to adaptable tools, fostering collaboration and innovation in the tech world.

Elon Musk to Claude Code

Anthropic has secured a partnership with Allianz, a major German insurance company, to integrate its AI technology into the insurance sector. The collaboration includes deploying Anthropic's AI coding tool, Claude Code, to Allianz employees, developing custom AI agents for workflow execution, and implementing a system to log AI interactions for transparency and regulatory compliance. Allianz CEO Oliver Bäte emphasized the partnership's focus on safety and transparency. This deal is part of Anthropic's recent string of enterprise agreements, including partnerships with Snowflake, Accenture, Deloitte, and IBM. Anthropic currently holds a significant share of the enterprise AI market, with a 40% overall market share and 54% in AI coding, as reported by Menlo Ventures.

From Claude Code to Albert Einstein's characters

At CES 2026, AI-enhanced collectibles are gaining attention with innovations like HeyMates and Buddyo. HeyMates, developed by Olli, integrates AI into Funko-like figurines using RFID chips, allowing users to interact with characters like Albert Einstein and fictional figures. Olli plans to launch these on Kickstarter, aiming to create a new market for AI-driven toys. Buddyo, led by Yijia Zhang, focuses on enhancing existing collectibles like Nintendo's Amiibo with an AI Pod that uses NFC technology. This platform allows users to create AI personalities for their figurines. Both companies aim to capitalize on the collectible figurine trend by integrating AI, despite Funko's current financial struggles. The market for AI toys and chatbot companions is poised for potential growth.

From collectibles, we shift to the missing Italian climber Nicola Ivaldo

AI technology played a crucial role in locating the body of missing Italian climber Nicola Ivaldo, who disappeared in September 2024. After initial search efforts failed, the Piemonte mountain rescue service employed AI to analyze thousands of drone-captured images of the Cottian Alps. The software identified a red helmet, leading rescuers to Ivaldo's body in a gully on Monviso's north wall. This marked a significant advancement in search and rescue operations, demonstrating AI's potential to expedite searches in challenging terrains. However, experts caution that AI still requires refinement to reduce false positives, especially in complex landscapes. Ethical considerations also arise regarding the use of aerial images in such operations, highlighting the need for responsible application.

Nicola Ivaldo fades as Experimental Advanced

China's Experimental Advanced Superconducting Tokamak (EAST), known as the "artificial sun," has achieved a significant milestone in nuclear fusion by maintaining plasma stability at extreme densities, surpassing the Greenwald Limit. This breakthrough, reported by the Chinese Academy of Sciences, suggests a potential pathway for extending density limits in fusion reactors, crucial for developing near-limitless clean energy. The EAST reactor managed to keep plasma stable at densities 1.3 to 1.65 times the Greenwald Limit, a challenge previously seen as a major obstacle. This advancement could inform future reactor designs, including the International Thermonuclear Experimental Reactor (ITER) in France, which aims to create sustained fusion reactions. Despite progress, nuclear fusion remains experimental, with reactors consuming more energy than they produce.

Experimental Advanced pivots to Isaac Lab

London-based startup Humanoid has rapidly developed its HMND 01 Alpha robots, moving from concept to prototype in seven months, significantly faster than the typical 18 to 24-month cycle. Utilizing NVIDIA's Isaac Lab and Isaac Sim, Humanoid employs a simulation-first approach, allowing for quick development and deployment of policies onto physical hardware. The robots, available in wheeled industrial and bipedal research versions, are undergoing field tests. Central to their design is the NVIDIA Jetson Thor platform, which simplifies system architecture and enhances serviceability. Humanoid aims to transition to software-defined standards, collaborating with NVIDIA on a new robotics networking system. Founded in 2024, the company has over 200 employees and reports 20,500 pre-orders with active pilot programs.

Isaac Lab then Eric Kauderer

Anthropic has launched new health care features for its Claude AI platform, allowing users to integrate and understand their medical data. This move follows OpenAI's introduction of ChatGPT Health, marking a significant push by AI companies into the health care sector. Both platforms enable users to share information from health records and fitness apps to personalize health discussions. However, there is ongoing scrutiny over AI's ability to safely interpret medical data. OpenAI's tool requires a waitlist, while Anthropic's offerings are available to Pro and Max subscribers in the U.S. Eric Kauderer-Abrams of Anthropic emphasized the potential of AI to simplify complex health care navigation. The Claude platform now includes a HIPAA-ready infrastructure, connecting to federal health care databases to assist providers. These features aim to automate tasks like preparing authorization requests and supporting insurance appeals. Despite the benefits, Anthropic and OpenAI caution that their systems are not substitutes for professional judgment, and human oversight remains essential. Anthropic’s policy mandates professional review for health-related decisions made using Claude.