Narus

China has introduced new AI regulations that will come into effect from September 2025. 🇨🇳 These new regulations, following a recent move by Spain, require all content created by AI to be clearly labelled, pointing to a global shift towards more transparency. Some key highlights from the new regulations: • All AI-generated content, including text, images, audio, video and virtual scenes, must have clear labels that are easy to spot and included in the file's background data. • In some cases, users can ask for content without labels, but this must be recorded properly. • It's now illegal to remove, hide or change these labels. • It's also banned to wrongly label human-made content as AI-generated. As more countries introduce similar measures, how should businesses respond to the growing push for AI transparency? #GenerativeAI #AIforBusiness #GenAI

Doors are open at DTX Manchester; let's do this! 📍 Meet Tim Sadler 🌍, Naia Arzanegui, and Adam Wignall at Booth C54 to discover how 𝐍𝐚𝐫𝐮𝐬 helps businesses get the most out of their GenAI tools. Check out the DTX 360 page for full event details. #DTXM25 #GenerativeAI #AIforBusiness #GenAI

AI isn't magic. It's only as powerful as the data behind it. Laura Gemmell, Founder of Taught by Humans, shares why more companies are now prioritising data quality before diving into artificial intelligence.⬇️ #GenerativeAI #AIforBusiness #GenAI

LLM costs can add up fast, but smart choices make a difference. Understanding where the costs come from is the first step. Usage, model selection, and optimisation all play a role, and without the right strategies, expenses can spiral. A few small tweaks can lead to big savings. Check out the tips below to keep costs in check while maximising performance. ⬇️ #GenerativeAI #AIforbusiness #GenAI

What happens when an AI model starts absorbing the emotional weight of their conversations? A recent study found that when large language models (LLMs) like ChatGPT are exposed to traumatic stories, their responses become more biased. Researchers observed that after being subjected to distressing prompts, ChatGPT displayed higher "anxiety" scores and generated increasingly skewed answers. This raises concerns, especially for AI-powered mental health chatbots like Woebot Health, which aim to provide cognitive behavioural therapy (CBT) support. Bias in AI is a known issue, but this research suggests that an AI's "emotional state" might influence the severity of its biases. The study tested whether mindfulness-based relaxation techniques could counteract this effect, and while they helped reduce AI "anxiety," they did not eliminate it entirely. If AI can be influenced by emotional context, should we manage its bias the same way we do with people? How do we ensure fairness in high-stakes areas like mental health and recruitment? Let's discuss. Source: https://lnkd.in/dUWMiXux #GenerativeAI #AIforbusiness #GenAI

Start your GenAI journey with Narus, your new safe-for-work AI chat tool ⚒️ Meet Tim Sadler 🌍 and his team at stand C54 and be entered into our competition to win a LEGO® set that is truly out of this world! #DTXM25

Some companies are overhyping their AI capabilities, whilst at the same time, employees are experimenting with tools without clear direction. Sue Turner OBE, Founding Director at AI Governance Limited, discusses what's really happening. ⬇️ #GenerativeAI #AIforbusiness #GenAI

RAG: have you heard of it? Retrieval-augmented generation (RAG) combines large language models with external data to produce more reliable and contextually relevant AI responses. Some benefits of RAG include: ✔️ Provides up-to-date, specific information from reliable external sources. ✔️ Improves the accuracy and relevance of AI-generated responses. ✔️ Reduces the risk of errors or generic outputs. RAG is also generally more cost-efficient and flexible than fine-tuning methods, as it doesn't require retraining the entire model. The future of RAG is set to revolutionise industries, driving smarter AI with real-time, precise data integration. #RAG #GenerativeAI #AIforbusiness #GenAI

AI is reshaping businesses. It has the power to transform operations, from customer service all the way through to risk management. Swipe through our examples of what is happening right now ➡️ ➡️ ➡️ ➡️ #GenerativeAI #AIforbusiness #GenAI

Super Mario Bros isn’t just a classic video game. 🎮 It's also an intensive AI training lab. Researchers at the University of California San Diego recently put AI models to the test using an emulated version of the game. The models had to generate Python code to control Mario, navigating obstacles and making split-second decisions. The results were unexpected. AI models designed for step-by-step reasoning struggled, while those optimised for speed performed better. In a game where precise timing is everything, slow decision-making proved to be a major disadvantage. So, how effective are these video games as benchmarks? • Success in games does not always reflect broader intelligence, as AI can be trained for specific tasks. • Critics argue that gaming benchmarks are more about publicity than real AI advancement. • AI models mastering a game like Mario doesn’t mean they can apply those skills elsewhere. Do video games provide meaningful insights into AI progress, or are they just a flashy distraction? #GenerativeAI #AIforbusiness #GenAI