What to expect from Google this week
Google I/O 2025 arrives with Google in third place in the foundation model race, particularly in coding capabilities. Despite this, the conference is expected to highlight Google's strengths in AI for science and potentially new developments in health AI, while also navigating internal controversies.
Google I/O 2025 commences with the company facing significant challenges in the foundation model arena, particularly concerning its coding capabilities. A year ago, Google was highly competitive with its Gemini 2.5 Pro, but now its coding tools reportedly lag behind those of Anthropic and OpenAI.
Reports suggest Google is seriously addressing its AI coding deficit, with a new team at DeepMind and the involvement of Nobel laureate John Jumper. However, it is unlikely Google will regain a leading position in AI coding during this conference, as internal reports indicate their publicly available models are still behind.
Despite coding weaknesses, Google DeepMind maintains a strong lead in AI for science, having earned a Nobel Prize for AlphaFold. Attendees should anticipate announcements regarding new scientific AI tools, such as improvements to its AI co-scientist or AlphaEvolve, building on its existing innovative work in the field.
Google also has substantial research in LLM-based health tools. While OpenAI's ChatGPT Health has recently dominated the conversation, Google plans to make its AI-powered Health Coach publicly available. It remains to be seen if this signals a new focus or a cautious approach in a high-stakes area.
Beyond product announcements, the conference will likely navigate recent controversies. A letter from 600 employees, many from DeepMind, protesting a DoD deal highlights internal dissent. Despite efforts to maintain neutrality, these issues may surface.
Related articles
Build real agentic apps using CUGA: two dozen working examples on a lightweight harness
CUGA, IBM's open-source Agent Harness, simplifies building agentic applications by handling infrastructure, allowing developers to focus on tools and prompts. It offers pre-assembled components for planning, execution, and state management, significantly reducing development time. CUGA has topped agent benchmarks like AppWorld and WebArena.
OpenAI launches new initiative to help find and patch open source bugs
OpenAI has launched "Patch the Planet," a new initiative in partnership with cybersecurity firm Trail of Bits, to enhance the security of open-source projects. This program aims to assist maintainers in identifying and patching bugs, utilizing OpenAI's AI-powered security tools while reducing the burden on project teams.
PP-OCRv6 on Hugging Face: 50-Language OCR from 1.5M to 34.5M Parameters
Baidu has released PP-OCRv6, an advanced optical character recognition (OCR) model supporting 50 languages. Available on Hugging Face, this version significantly improves accuracy and efficiency across various parameter sizes, from 1.5 million to 34.5 million, marking a substantial leap in multilingual OCR technology.
