Nuclear energy could play a role in meeting the high-power demands of AI systems, especially for continuous and stable power. Microsoft is exploring small modular nuclear reactors (SMRs) for data centers hosting its cloud and AI services, and is seeking a program manager to develop a strategy.
"This senior position is tasked with leading the technical assessment for the integration of SMR and microreactors to power the data centers that the Microsoft Cloud and AI reside on."
Critics of nuclear power worry about safety, waste storage, and competition with renewable energy sources.
Just a quick FYI: You can now preview DALLE-3 in Bing Image Creator before the full chat version launches in ChatGPT Plus in October. The prompt for the image below was "Exterior view of an old junk store in New York in the 80s. The shop sign reads "THE DECODER". In the window of the store is a sign that says "AI NEWS FOR FREE". I used the Image Creator via Edge.
Image: DALL-E 3 prompted by THE DECODER
The result seems great, much better than existing systems in following the prompt, but not perfect when it comes to the writing. Let's see how good the ChatGPT version works, since it offers the option to refine images. The image above is the best one out of four.
Google introduced Google-Extended, a new control that allows web publishers to manage how their content is used for generative AI use cases such as Bard and Vertex AI, as well as "future generations of models." The feature provides more choice and control over website content, Google said. "A website administrator can choose whether to help these AI models become more accurate and capable over time." Similar to OpenAI's GPTBot, you can use robots.txt to block Google-Extended.
User-agent: Google-Extended Disallow: /
It's not clear from the announcement if blocking Google-Extended will cause Bard and SGE to stop using your content in chat, or if it's just about training data.
More than 20 girls in the Spanish town of Almendralejo received AI-generated nude photos of themselves, sparking debate over whether deepfakes can be legally punished, Euronews reports. The images were reportedly altered using the app ClothOff ("Undress AI, Undress girl for free!") and shared in WhatsApp groups. The images don't show the girls' real bodies, but are still causing great distress to the victims, some of whom are as young as 11. The national police have launched an investigation and authorities are considering whether the images in circulation could be considered child pornography.
