At the recent 'Invest Karnataka 2025' conference held in Bangalore, numerous leaders from the technology industry gathered to discuss the transformative potential of artificial intelligence in India and its impacts. Manish Gupta, Senior Director at Google DeepMind, delivered a key speech at the conference, emphasizing the need to establish corresponding regulations alongside technological innovation to ensure responsible development. Note: Image generated by AI, image courtesy service provider Midjourney.

Recently, former DeepMind scientist Simon Kohl announced the establishment of a new artificial intelligence drug discovery company, aiming to leverage his extensive experience in protein prediction, particularly related to the Nobel Prize-winning AlphaFold system. AlphaFold is a revolutionary AI tool that can accurately predict the three-dimensional structure of proteins, a breakthrough that has garnered significant attention in biomedical research. Image source note: Image generated by AI, image licensed service.

Recently, an AI system developed by Google DeepMind—AlphaGeometry2—has successfully surpassed the average level of gold medalists at the International Mathematics Olympiad (IMO), demonstrating outstanding performance in solving geometry problems. AlphaGeometry2 is an upgraded version of the AlphaGeometry system released by DeepMind last year. According to the latest research from the research team, this system can solve 84% of the geometry problems from the IMO over the past 25 years.

At the recent World Economic Forum in Davos, Google DeepMind's CEO Demis Hassabis stated that the first drugs designed with the help of artificial intelligence could begin clinical trials by 2025. Hassabis also heads Isomorphic Labs, a drug research company under DeepMind. He stated, "Our plan is to have some AI-designed drugs enter clinical trials by the end of this year."

In multimodal tasks, visual language models (VLMs) play a crucial role, such as in image retrieval, image captioning, and medical diagnosis. These models aim to align visual data with language data for more efficient information processing. However, current VLMs still face significant challenges in understanding negation. Negation is critical in many applications, such as distinguishing between "a room without windows" and "a room with windows." Despite significant progress made by VLMs, existing models fall short when it comes to handling negated statements.

In terms of artificial intelligence (AI) regulation, the UK government has expressed a desire to take an independent stance and plans to adopt different approaches from its main Western counterparts, such as the EU and the US. UK AI and Digital Government Minister Feryal Clark emphasized in an interview with CNBC that the UK must 'do its own thing' to ensure necessary regulations are in place early on regarding the safety of AI models. Image source note: The image was generated by AI and provided by the image service provider Midjourney.

In the AI field, talent mobility has always been a fiercely competitive arena. Recently, the job switch of Tim Brooks has drawn widespread attention. After working at Sora for only three months, he announced his move to the competitor Google DeepMind. This move is seen as a response to serious challenges faced by Sora's technology, particularly in generating speed and performance, which have been underwhelming compared to other players in the industry. Reports indicate that Sora has numerous technical issues and struggles to compete with Luma and Stabilit.

Google recently announced the formation of a new team dedicated to developing artificial intelligence (AI) models capable of simulating the physical world. The team will be led by Tim Brooks, who previously co-led the video generation model Sora at OpenAI. Brooks stated on social media platform X that this new team will belong to Google's AI research lab, Google DeepMind. In his announcement, Brooks mentioned, 'DeepMind's plans are ambitious.'

Recently, Felix Hill, a research scientist at Google DeepMind, tragically passed away at the age of 41 due to mental health issues. This news has sparked widespread attention and mourning within the AI community. Felix had been struggling with severe mental health problems since early 2023, exhibiting significant depression and suicidal tendencies. Despite his notable achievements in academia, he ultimately could not overcome his internal struggles. Felix Hill was an outstanding AI scholar who contributed to the creation of natural language processing.

Generative language models face many challenges during the transition from training to practical applications. One major issue is how to achieve optimal performance during the inference stage. Current strategies, such as Reinforcement Learning from Human Feedback (RLHF), primarily focus on improving the model's win rate but often overlook decoding strategies during inference, such as Best-of-N sampling and controlled decoding. This gap between training objectives and actual usage may lead to inefficiencies and affect the quality and reliability of outputs. To address these issues, Google D

Recently, Google's deep learning team, in collaboration with researchers from several universities, announced a new system called 'MegaSaM.' This system can quickly and accurately estimate camera parameters and depth maps from regular dynamic videos. The advent of this technology will open up more possibilities for the videos we record in our daily lives, particularly in capturing and analyzing dynamic scenes. Traditional Structure from Motion (SfM) and monocular Simultaneous Localization and Mapping (SLAM) techniques.

Google DeepMind and robotics company Apptronik have announced a collaboration to combine their technical expertise in developing AI humanoid robots capable of operating in complex dynamic environments. This partnership aims to advance the deep integration of robotic hardware and artificial intelligence, exploring broader possibilities for robots in both industrial and everyday applications. Apptronik has been independently operating since 2016, spun off from the Human Centered Robotics Lab at the University of Texas at Austin, and has developed 15 different robot models to date, including the...