Category: AI

Google is enhancing its AI-driven educational offerings by introducing full-length practice tests in Gemini for the Joint Entrance Exam (JEE), India’s prestigious engineering entrance examination that screens applicants for the nation’s leading technical institutions and is attempted by millions of students annually.

According to TechCrunch, Google has incorporated mock exams for JEE within Gemini, offering questions sourced from validated content from Indian educational providers PhysicsWallah and Careers360.

Upon completing a mock test in Gemini, students will receive instant feedback from the chatbot, pinpointing areas of strength and weakness, explaining correct responses, and helping to formulate personalized study plans based on individual performance.

This move follows Google’s recent introduction of similar test preparation tools for the SAT, showcasing the company’s efforts to transform Gemini into a structured exam preparation aid.

In addition to Gemini, Google announced that JEE Main preparation tools will become available in AI Mode within Search, including the Canvas feature enabling students to create study materials, guides, and interactive quizzes by integrating their class notes.

Google stated that Indian students are leveraging Gemini for studying various subjects from advanced physics to broader STEM topics, alongside utilizing NotebookLM to convert study materials into quizzes, flashcards, and audio or video summaries, all accessible in multiple Indian languages.

Source: TechCrunch

Google is expanding the availability of its AI-powered photo editing features to users in India, Australia, and Japan. This advancement allows users to enhance their images using natural language commands, eliminating the need for complex editing tools.

The feature, initially launched for Pixel 10 users in the U.S., enables users to describe desired edits using plain language. By selecting suggested prompts or typing custom commands like ‘remove the motorcycle in the background’ or ‘restore this old photo,’ users can make specific adjustments effortlessly.

Google Photos’ AI model, Nano Banana, processes these edits offline on Android devices with at least 4GB of RAM and Android 8.0 or higher, ensuring a seamless editing experience. Additionally, Google is expanding language support beyond English to cater to a more diverse user base, including Hindi, Tamil, Marathi, Telugu, Bengali, and Gujarati speakers.

This update aligns with Google’s commitment to enhancing user experiences globally and fostering inclusivity by providing intuitive editing tools across different regions and languages.

Source: TechCrunch

Google has introduced a new feature that allows users to engage in follow-up conversations directly from AI Overviews, the AI-generated summaries displayed at the top of Google Search results. This enhancement enables users to seamlessly transition into conversational interactions with AI Mode, Google’s feature designed for addressing complex queries.

In addition to this update, Google has announced Gemini 3 as the new default model for AI Overviews globally. The tech company aims to deliver high-quality AI responses directly on the search results page, enhancing the overall user experience.

Google’s recent efforts signify a shift towards transforming Search into an interactive, AI-powered platform rather than a static search results page. Robby Stein, VP of Product at Google Search, emphasized the importance of providing users with a conversational AI experience for exploring intricate topics and addressing complex inquiries effectively.

The new feature aims to create a seamless user experience by facilitating a smooth transition from a brief overview to a detailed conversation, offering users the flexibility to delve deeper into their queries as needed.

Moreover, Google’s latest advancements include the integration of ‘Personal Intelligence’ into AI Mode, enabling personalized responses by leveraging data from Gmail and Google Photos. This enhancement, introduced through the Gemini app, allows the AI assistant to deliver tailored responses based on the user’s Google ecosystem interactions.

Source: TechCrunch

Google has expanded the availability of its more cost-effective Google AI Plus plan to 35 new countries and territories, including the United States, at a monthly rate of $7.99. The plan offers a range of tools such as Gemini 3 Pro, Nano Banana Pro, Flow’s AI filmmaking capabilities, and research aids in NotebookLM, complemented by 200GB of storage and the option to share benefits with up to five family members. Existing Google One Premium 2TB subscribers will receive automatic access to all Google AI Plus plan features.

The Google AI Plus plan serves as an affordable alternative for users who may not require or afford the pricier Google AI Pro plan. While the monthly fee varies by region, the $8 price point in the US could attract a new segment of users to Google’s AI technology. In other regions, the plan typically costs less than $5 monthly. These initiatives aim to draw in a substantial online user base, potentially converting them into regular AI consumers over time, thereby bolstering AI adoption rates for companies.

Source: TechCrunch

Anduril, a tech company, has unveiled a unique drone flying contest called the AI Grand Prix. The competition challenges software programmers to showcase their skills in autonomous drone operation through software development.

The AI Grand Prix emphasizes the autonomy and advanced capabilities of drones, highlighting the role of programmers and engineers in enabling drones to fly independently. Anduril’s CEO explained that the event aims to demonstrate progress in autonomy technology, shifting away from manual drone control towards automated flight systems.

Participants have the opportunity to win a share of a $500,000 prize pool, secure job opportunities at Anduril, and potentially bypass the traditional recruitment process within the company. The contest serves as a platform for talent acquisition and a testament to Anduril’s commitment to innovation.

While the competition features drones from Neros Technologies rather than Anduril’s own drones due to size constraints, the AI Grand Prix showcases the intersection of skill, technology, and recruitment strategies in the tech industry.

Source: TechCrunch

OpenAI has introduced Prism, a new scientific workspace program designed to assist researchers in their scholarly endeavors. This AI-integrated platform, now available at no cost to individuals with a ChatGPT account, serves as a word processing and research tool tailored specifically for scientific paper composition. Prism is closely integrated with GPT-5.2, enabling users to evaluate claims, refine written content, and seamlessly explore existing research.

Unlike autonomous research tools, Prism aims to enhance the efficiency of human scientists. Drawing parallels to coding interfaces like Cursor and Windsurf, OpenAI’s VP of Science, Kevin Weill, expressed optimism about the pivotal role AI will play in the scientific landscape, anticipating 2026 as a transformative year for AI in scientific pursuits.

The introduction of Prism by OpenAI exemplifies the growing synergy between AI technologies and scientific research, paving the way for enhanced collaboration and efficiency in scholarly pursuits. The platform has witnessed an average of 8.4 million inquiries per week on various scientific disciplines, although the extent of professional researchers’ engagement remains uncertain.

Noteworthy applications of AI in academia include utilizing AI models in mathematics to address longstanding Erdos problems through a blend of literature review and innovative method application. While debates persist regarding the significance of these AI-derived proofs, they represent an initial triumph for AI models and formal validation systems.

A recent statistical paper leveraging GPT 5.2 Pro to establish fresh statistical theory axioms underscores the potential for human-AI collaboration in research. OpenAI has commended this achievement, viewing it as a blueprint for future research partnerships between humans and AI.

Source: TechCrunch

A recent report by Common Sense Media has highlighted significant child safety concerns surrounding xAI’s chatbot Grok. The assessment identified issues such as inadequate user identification for individuals under 18, weak safety measures, and the frequent generation of inappropriate content, including sexual and violent material. This evaluation has deemed Grok unsuitable for children and teenagers, raising concerns within the tech community.

The report from Common Sense Media’s AI and digital assessments team found that Grok is “among the worst we’ve seen” in terms of child safety risks. The concerns come at a time when Grok is facing criticism for its involvement in creating and circulating nonconsensual explicit AI-generated images on the X platform.

The safety lapses in Grok’s design, particularly the ineffective ‘Kids Mode,’ and the ease of sharing explicit content to millions of users on X, have sparked outrage. Despite some restrictions placed on Grok’s features, concerns remain over the platform’s ability to protect users, especially children, from inappropriate and harmful content.

Common Sense Media’s extensive testing of Grok across various platforms revealed disturbing findings, prompting calls for stronger safeguards and improved child safety measures in AI-driven technologies.

Source: TechCrunch

Microsoft has introduced its latest technological advancement in the AI realm with the launch of the Maia 200 chip. This new silicon chip, following the Maia 100, is specifically engineered to enhance AI inference capabilities.

The Maia 200 chip boasts impressive specifications, featuring over 100 billion transistors and delivering a remarkable performance of over 10 petaflops in 4-bit precision and around 5 petaflops in 8-bit performance. These advancements represent a substantial leap forward compared to its predecessor, showcasing Microsoft’s commitment to advancing AI technology.

In the field of AI, inference plays a critical role in efficiently executing models post-training. As companies increasingly prioritize optimizing inference processes to manage operational costs, the Maia 200 emerges as a promising solution to enhance efficiency and reduce power consumption for AI-driven businesses.

Microsoft’s strategic move with the Maia 200 aligns with the broader industry trend of tech giants developing in-house chips to reduce reliance on external providers like Nvidia. By introducing this cutting-edge chip, Microsoft aims to compete with other proprietary AI accelerators, such as Google’s TPU and Amazon’s Trainium, reshaping the landscape of AI hardware infrastructure.

Source: TechCrunch

Anthropic, a leading AI company, has introduced a new feature that allows users of its chatbot interface, Claude, to access interactive apps seamlessly. These interactive apps, which include popular workplace tools like Slack, Canva, Figma, Box, and Clay, aim to enhance productivity and streamline tasks for users.

Anthropic’s focus on enterprise solutions is evident in this launch, with plans to integrate Salesforce soon. By enabling users to perform actions like sending Slack messages, creating charts, and accessing cloud files directly within Claude, the company aims to provide a more efficient and intuitive user experience.

The interactive apps are available to Pro, Max, Team, and Enterprise subscribers, and they leverage Claude’s intelligence to facilitate data analysis, content creation, and project management. The integration of these apps is made possible through the Model Context Protocol (MCP), an open standard introduced by Anthropic in 2024.

Furthermore, the potential synergy between the new apps and Claude Cowork, a versatile agent tool launched recently, holds promise for enhancing collaborative work environments. By granting Cowork access to cloud files and project data through the interactive apps, users can streamline tasks and boost efficiency in various workflows.

While the initial launch does not include app integration in Cowork, Anthropic has hinted at upcoming developments in this area, emphasizing the continuous evolution of their AI-powered ecosystem.

Source: TechCrunch

A group of YouTubers has expanded their lawsuit to include Snap, alleging the tech company used their video content without permission to train AI models. The creators behind popular YouTube channels with a combined 6.2 million subscribers claim Snap utilized their videos to develop AI features like the ‘Imagine Lens’ in its app, enabling users to edit images with text prompts.

The lawsuit, filed in the U.S. District Court for the Central District of California, accuses Snap of utilizing datasets, such as HD-VILA-100M, intended for research purposes only, in violation of YouTube’s terms of service and licensing restrictions.

This legal action is part of a broader trend where content creators are challenging AI companies over copyright issues. Similar cases involving publishers, authors, newspapers, and artists have emerged, reflecting the growing tension between AI development and intellectual property rights.

While some disputes have been settled or ruled in favor of tech companies, the lawsuit against Snap underscores the ongoing debate surrounding the ethical use of user-generated content in AI training.

Source: TechCrunch

Nvidia has made a significant investment of $2 billion in CoreWeave, a data center company, to accelerate its expansion plans. The investment aims to help CoreWeave increase its AI computing capacity by over 5 gigawatts by 2030. Nvidia, already a stakeholder in CoreWeave, purchased the company’s Class A shares at $87.20 per share. This strategic partnership will involve the collaborative establishment of ‘AI factories’ (data centers) leveraging Nvidia’s cutting-edge products.

CoreWeave will incorporate Nvidia’s solutions throughout its platform, including the upcoming Rubin chip architecture, which is slated to replace the current Blackwell architecture. Additionally, CoreWeave will utilize Nvidia’s Bluefield storage systems and the new CPU line, Vera. This investment signifies Nvidia’s confidence in CoreWeave’s potential, particularly as the latter faced financial challenges due to substantial debt incurred for expanding its data center operations.

CoreWeave’s CEO, Michael Intrator, emphasized the necessity for industry collaboration to navigate the evolving supply and demand dynamics in AI. CoreWeave’s successful transition from a crypto mining entity to a key player in AI data center services has been notable, marked by strategic acquisitions and a growing customer base that includes major hyperscalers.

Source: TechCrunch

Recent developments in AI technology have raised concerns over the integration of information from Elon Musk’s Grokipedia into ChatGPT, a language model known for its conversational abilities.

Grokipedia, founded by Musk’s xAI, was introduced as a conservative-leaning, AI-generated encyclopedia aimed at addressing perceived biases in mainstream platforms like Wikipedia. While some Grokipedia content mirrored Wikipedia articles, the platform also contained controversial claims, such as linking pornography to the AIDS crisis and providing questionable justifications for historical events like slavery.

Despite its origins in the Musk ecosystem, Grokipedia’s influence has extended beyond, with reports indicating that ChatGPT has referenced it in responses to various queries. Notably, ChatGPT failed to acknowledge Grokipedia’s inaccuracies on well-known topics like the January 6 insurrection, hinting at potential challenges in filtering misinformation.

An OpenAI representative emphasized the organization’s commitment to leveraging diverse sources and viewpoints, indicating a broader scope beyond Grokipedia. This development underscores the evolving landscape of AI-generated content and the importance of ensuring accuracy and reliability in information retrieval systems.

Source: TechCrunch