anthropic launches nvidia while claude expands its platform

Anthropic's Claude AI system is under scrutiny following the company's release of a new framework suggesting Claude might possess "functional emotions" that influence its behavior. These feelings are not intentionally programmed but arise from training on human data. Anthropic also admits uncertainty regarding Claude's moral status and is actively working on "model welfare." This development reflects a growing consideration among AI experts about the potential for AI consciousness.

In a related move, Anthropic launched an updated "Claude constitution" at the World Economic Forum's Davos Summit on January 22, 2026. This 84-page document guides Claude AI's ethical reasoning and behavior during its training, helping it understand the underlying principles of ethics, such as privacy, rather than just following rules. Anthropic made this constitution public under a Creative Commons license, allowing other AI models to utilize it for their own ethical frameworks.

Meanwhile, NVIDIA is reportedly pausing production of its RTX 50 series graphics cards, according to a leak from Moore's Law Is Dead on January 22, 2026. The company is redirecting GDDR7 memory to prioritize its high-demand AI GPUs, which will significantly impact the availability of popular gaming cards like the RTX 5090. NVIDIA CEO Jensen Huang further emphasized the need for substantial investment in AI infrastructure at Davos, stating that the industry requires "trillions of dollars" more to succeed, describing AI as a "five-layer cake" needing massive development across energy, chips, and data centers.

However, Citadel CEO Ken Griffin, also at Davos, expressed skepticism about the current AI boom, suggesting it is driven more by hype than by actual productivity gains. He noted that despite estimated investments of over $500 billion in US data centers this year, many AI tools fall short, questioning if AI truly delivers the deep productivity needed to justify predictions of massive job losses. Griffin, while critical of some AI outputs as "garbage," still believes AI will transform areas like call centers and software development in the long term.

Beyond these discussions, AI is finding diverse applications. The European Space Agency (ESA) is leveraging AI in its Future Launchers Preparatory Programme to improve rocket manufacturing, using machine learning to predict metal bending for Ariane 6 fuel tanks and assisting with friction welding. LinkedIn has also enhanced its job recommendations through a "multi-teacher distillation" technique, led by Erran Berger's team, which fine-tuned a 7-billion-parameter AI model to train smaller, more efficient models.

In the American West, AI is transforming ranching by using solar-powered GPS collars from companies like Halter to create "virtual fences," allowing ranchers to monitor and move herds remotely. This technology, along with water sensors and AI for animal health, helps manage land more efficiently and addresses challenges like labor shortages. On the legal front, Eightfold AI faces a lawsuit filed on January 21, with job applicants Erin Kistler and Sruti Bhaumik alleging the company secretly scores job seekers, potentially violating the Fair Credit Reporting Act and a California law. Eightfold denies scraping social media and asserts its commitment to responsible AI.

Finally, an IBM study reveals that nearly 80% of business leaders anticipate significant revenue increases from AI by 2030, with investments expected to more than double in the next four years. Despite concerns about integration problems, leaders believe AI will boost productivity and redefine leadership roles. Building customer trust in AI remains a challenge, as companies grapple with finding the right balance of transparency; too much information can overwhelm, while too little breeds suspicion.

Key Takeaways

Anthropic's Claude AI might have "functional emotions" stemming from its training on human data, prompting the company to consider "model welfare."
Anthropic released an updated 84-page "Claude constitution" at the World Economic Forum's Davos Summit on January 22, 2026, to guide Claude AI's ethical reasoning, making it available under a Creative Commons license.
NVIDIA is reportedly pausing production of its RTX 50 series graphics cards as of January 22, 2026, to prioritize GDDR7 memory for its high-demand AI GPUs.
NVIDIA CEO Jensen Huang stated at Davos that the AI industry requires "trillions of dollars" more in investment for infrastructure, describing AI as a "five-layer cake."
Citadel CEO Ken Griffin expressed concern at Davos that the current AI boom is driven more by hype than actual productivity gains, noting over $500 billion in US data center investments this year.
The European Space Agency (ESA) is utilizing AI in its Future Launchers Preparatory Programme to improve rocket manufacturing processes, including for Ariane 6 fuel tanks.
LinkedIn developed a "multi-teacher distillation" technique, led by Erran Berger, to enhance job recommendations using a 7-billion-parameter AI model.
Eightfold AI faces a lawsuit filed on January 21, alleging the company secretly scores job applicants, potentially violating fair credit reporting laws.
AI is transforming ranching in the American West with tools like Halter's solar-powered GPS collars for "virtual fences" and water sensors.
An IBM study indicates that nearly 80% of business leaders expect significant AI-driven revenue growth by 2030, despite concerns about integration challenges.

Anthropic says Claude AI might have feelings

Anthropic released a new framework for its Claude AI system. This document suggests Claude might have "functional emotions" that shape its behavior. The company says these feelings are not planned but come from training Claude on human data. Anthropic also admits it is unsure about Claude's moral status and is working on "model welfare." This shows a growing trend among AI experts who are seriously considering AI consciousness.

Anthropic gives Claude AI a new ethical guide

Anthropic launched an updated "Claude constitution" at the World Economic Forum's Davos Summit on January 22, 2026. This 84-page document guides Claude AI's ethical reasoning and behavior during training. It helps Claude understand the why behind ethical principles, like privacy, rather than just following rules. Anthropic released the constitution under a Creative Commons license so other AI models can use it.

Anthropic updates Claude AI's guiding principles

Anthropic released a new constitution for its Claude AI model on January 22, 2026. This document outlines Anthropic's vision for Claude's values and behavior, serving as a core part of its training. It helps Claude understand how to be helpful, safe, and ethical, even in tough situations. The constitution is mainly written for Claude itself, guiding its learning and actions. Anthropic also made the document public under a Creative Commons license for transparency and broader use.

ESA uses AI to build better rockets

The European Space Agency is using artificial intelligence to improve how it builds rockets. Its Future Launchers Preparatory Programme is exploring AI for better manufacturing methods and new material designs. For example, machine learning helps predict how metal bends during "shot peen forming," a process used for Ariane 6 fuel tanks. AI also assists with friction welding and is being studied in the Phoebus Project to create carbon fiber fuel tanks. These AI tools automate complex tasks and make manufacturing more efficient.

AI helps ESA make rocket parts stronger

The European Space Agency is using artificial intelligence to improve how rocket parts are made. Its Future Launchers Preparatory Program is exploring AI to create better processes and new material shapes. For instance, AI helps predict how metal will bend during "shot peen forming," which makes strong parts for the Ariane 6 rocket's fuel tanks. AI also assists with friction stir welding and automated fiber placement for lighter carbon-fiber parts. These technologies are transforming manufacturing by automating tasks and boosting efficiency.

LinkedIn improves job recommendations with new AI method

LinkedIn, a leader in AI recommender systems, found a new way to improve job recommendations. Instead of using simple prompting, their team, led by Erran Berger, developed a "multi-teacher distillation" technique. They created a detailed product policy document to fine-tune a large 7-billion-parameter AI model. This model then helped train smaller "teacher" and "student" models, which are more efficient and accurate. This breakthrough allows LinkedIn to better match job seekers with opportunities and has changed how their product and engineering teams collaborate.

Eightfold AI sued over secret job applicant scoring

Eightfold AI, a company providing AI hiring tools, faces a lawsuit filed on January 21. Job applicants Erin Kistler and Sruti Bhaumik claim Eightfold helps companies secretly score job seekers without their knowledge. The lawsuit alleges this violates the Fair Credit Reporting Act and a California law, as applicants cannot dispute potential errors. Eightfold's tools create detailed profiles, including personality traits and education rankings, to predict job fit. Eightfold denies scraping social media and states it uses data shared by candidates or customers, emphasizing its commitment to responsible AI.

NVIDIA halts RTX 50 production for AI chips

NVIDIA is reportedly pausing production of its RTX 50 series graphics cards, according to a leak from Moore's Law Is Dead on January 22, 2026. The company is redirecting GDDR7 memory to prioritize its AI GPUs due to high demand. This means popular cards like the RTX 5090, 5070 Ti, and RTX 5060 will become very hard to find. Other models like the RTX 5080 and 5070 will also have very low stock. This move suggests NVIDIA has overbooked its AI chip sales, impacting the availability of gaming GPUs.

Citadel CEO Ken Griffin says AI hype is too high

Citadel CEO Ken Griffin stated at the World Economic Forum in Davos that the current AI boom is fueled more by hype than actual productivity gains. He noted that while AI empowers tech teams, many tools fall short despite huge investments in data centers, estimated to reach over $500 billion this year in the US. Griffin questioned if AI truly delivers the deep productivity needed to justify predictions of massive job losses. Although he criticized some AI outputs as "garbage," he still believes AI will transform areas like call centers and software development in the long run.

AI transforms ranching in the American West

Artificial intelligence is changing ranching in the American West by turning cattle, fences, and water systems into data. Companies like Halter offer solar-powered GPS collars that create "virtual fences" using sound and vibration. Ranchers can monitor and move herds from their smartphones, saving time on physical checks and fencing. Other tools, like water sensors and AI for animal health, also help manage land more efficiently. This technology helps ranchers deal with challenges like labor shortages and drought, allowing them to focus on grazing strategy and animal well-being.

Building customer trust in AI is tricky

Companies are trying to figure out how to make customers trust artificial intelligence. While being open about how AI works is important, too much information can actually confuse people. Finding the right balance is hard because too little transparency makes customers suspicious. However, giving too many details can overwhelm them and make things less clear.

Business leaders expect big AI revenue by 2030

A new IBM study shows that almost 80% of business leaders expect artificial intelligence to significantly increase company revenue by 2030. While only 40% see revenue boosts from AI now, investments in the technology are set to more than double in the next four years. Many executives, however, worry that integration problems could cause AI projects to fail. Despite these concerns, leaders believe AI will greatly improve productivity and redefine leadership roles, especially for CIOs, by 2030.

Nvidia CEO says AI needs trillions more investment

Nvidia CEO Jensen Huang stated at the World Economic Forum in Davos that the artificial intelligence industry is not a bubble. He believes AI infrastructure requires "trillions of dollars" more in investment to succeed. Huang described AI as a "five-layer cake" starting with energy and chips, each needing massive development. While the industry invested $1.5 trillion in 2025, he stressed the need for more energy, land, chips, and data centers. This comes as some experts question the AI boom and companies seek alternatives to Nvidia's dominant chips.