March 19 AI news: Nvidia chip, European legislation, Apple and GPT-5
What an exciting start to the week in the world of AI! Nvidia made headlines at its annual conference in San Jose, unveiling innovations set to revolutionize the sector. Meanwhile, Europe is making strides in regulating this technology with the adoption of unprecedented legislation by the European Parliament. Let’s take a look at the most significant AI news from the last 48 hours.
Nvidia introduces its “superchip” Blackwell
The American graphics card manufacturer took advantage of its annual keynote, the GTC, to unveil its new AI-dedicated chip architecture with great fanfare. Named Blackwell, this family of GPUs promises performances up to 4 times greater than the previous generation while being up to 25 times more energy-efficient. This could significantly speed up the training of AI models, which require extensive computational power.
Europe adopts pioneering legislation on AI
In a global first, the European Parliament voted on Wednesday for a binding legal framework to regulate the use of artificial intelligence. The goal is to protect citizens’ rights and safety while promoting innovation. The text notably plans to ban social scoring systems and mass surveillance. High-risk AIs, such as those used in law enforcement, will be subject to strict requirements.
Apple negotiates with Google to integrate its AI Gemini
The Cupertino firm is reportedly in advanced discussions with Google to integrate its conversational AI, Gemini, into the next version of iOS, the operating system of iPhones. This move would allow Apple to catch up in this area against competitors like Microsoft, Meta, or Amazon. The tech giant has also approached OpenAI, the creator of ChatGPT.
GPT 4.5 and GPT 5 in the pipeline?
While the ink on the release of GPT-4 is barely dry, persistent rumors are circulating about the imminent arrival of even more powerful versions of OpenAI’s famous language model. In a recent interview, CEO Sam Altman mentioned the existence of GPT 4.5 and even GPT 5, without giving a specific release date.
These future iterations could incorporate multimodal capabilities (text, image, sound) and achieve a level of performance close to human intelligence on certain tasks. This fuels both fantasies and fears about general AI (AGI), whose specter grows closer by the day.
Stable AI Revolutionizes 3D with Stable Video 3D
Stability AI, the company behind the famous Stable Diffusion image generation model, has taken a significant leap forward with Stable Video 3D. This new AI can generate photorealistic 3D videos from a single image by synthesizing new views of the object from different angles.
Available in two versions (SV3D_u and SV3D_p), this model significantly outperforms open-source alternatives like Zero123-XL, both in terms of quality and consistency of the generated views. Stable Video 3D is already available for commercial use to Stability AI members, while the model weights can be downloaded from Hugging Face for non-commercial use.
📌As you can see, the AI news is particularly dense at the moment, with announcements coming at a frantic pace. No doubt the coming days will still hold their share of surprises.
