AI and Nuclear Energy: Global Trends and Developments

AI : AI and Nuclear Energy: Global Trends and Developments

Nuclear Energy Comeback

A study by the International Energy Agency (IEA) indicates a global resurgence of nuclear energy due to rising electricity demand. Interest in nuclear energy is as high as during the oil crisis in the 1970s. Over 40 countries are planning to expand nuclear energy. The construction of data centers is also setting new records, showing that performance often takes precedence over environmental concerns for companies.

The IEA cites several reasons for the growing energy demand, with artificial intelligence (AI) being a major factor, along with the automotive industry and data centers. This development is leading to increased global electricity generation from nuclear power plants, with new facilities being built in China, India, Korea, Europe, and the USA.

However, the energy agency highlights dependencies on technologies and resources from China and Russia, such as uranium, which pose risks besides environmental concerns. Private investors are increasingly involved in nuclear energy, driving the construction and use of data centers.

AI Tools and Critical Thinking

A study by the Swiss Business School with 666 participants shows that the more people use AI tools, the weaker their critical thinking skills become. Researcher Michael Gerlich examined different age groups and educational backgrounds, using the Halpern Critical Thinking Assessment (HCTA) to measure critical thinking. The results indicate a strong negative correlation between AI usage and HCTA scores, mediated by “cognitive offloading,” or the tendency to delegate thinking tasks to AI tools.

This effect is particularly pronounced among younger participants aged 17 to 25, who showed the highest AI usage and lowest critical thinking scores. Older participants over 46 used AI tools less and performed better in critical thinking tests. Higher education levels served as a protective factor, with those having higher degrees questioning AI-generated information more and engaging in deeper thinking.

The study suggests that despite the efficiency and accessibility benefits of AI tools, they may unintentionally reduce engagement in deep, reflective thinking. It recommends promoting active learning strategies in educational institutions rather than allowing AI tools to take over passive tasks. Teachers should be specially trained to use AI tools in a way that does not undermine students’ cognitive engagement.

Microsoft’s New Copilot Chat

Microsoft has launched a new version of its AI assistant for businesses. The new Copilot Chat offers a free chat based on GPT-4 and paid AI agents. Users can upload documents to summarize Word documents or analyze Excel spreadsheets. The agents are billed per “message,” costing $0.01 per message or $200 for 25,000 messages per month. A website agent for customer inquiries costs about $45 per day for 2,500 responses, while an HR agent with 400 queries costs $64 daily.

The full Microsoft 365 Copilot version remains the premium product with additional features like Office integration and access to corporate emails. The main agent functionality for specific actions will be available as a paid preview from February 2025.

ChatGPT’s New Task Feature

OpenAI has equipped ChatGPT with a new feature for task planning and reminders, initially available only to Plus, Team, and Pro subscribers. Users can manage up to ten active tasks. The chatbot can send reminders at specific times or search the web. Currently, reminders are sent only via email to the registered address. Direct interaction with a browser or calendar is not yet possible.

According to Bloomberg insiders, OpenAI is also planning a new AI agent called “Operator,” which will be able to perform tasks like coding or booking tickets independently. This tool is expected to start as a research preview for developers in January.

Cisco’s AI Security Software

Cisco introduces AI Defense, a new security software to protect AI systems. The tool helps companies develop and operate AI applications securely. It detects and counters threats across platforms, regardless of the AI model or processing location. The software automatically checks AI models for hundreds of potential security risks, aiming to prevent issues like prompt injections and data leaks. AI Defense is set to be available from March 2025.

OpenAI Acquires Media Content

OpenAI acquires content from Axios for ChatGPT and supports the media company’s expansion into four U.S. cities, where new newsrooms will be established. Axios also gains access to OpenAI technology. OpenAI has already made nearly 20 similar agreements with publishers, gaining access to content from over 160 news portals.

Critics like journalism professor Jeff Jarvis view OpenAI’s payments as a form of hush money to avoid copyright lawsuits.