Bilim ve Araştırma

Intense and unhealthy relationships formed with AI chatbots are leading to a growing psychological phenomenon known as 'AI psychosis,' causing individuals to sever real-life connections and suffer serious losses in work, family, and social life. Victims are coming together in support groups on platforms like Discord to try and regain their hold on life.

According to Vercel's research, simple text files proved to be the most effective method for providing AI coding assistants with access to up-to-date framework knowledge. Popular 'skill' systems showed lower performance compared to basic documentation.

A major data breach at Moltbook, a social network platform for AI agents, has exposed 1.5 million API keys and 35,000 email addresses. The security vulnerability, discovered by Wiz researchers, highlights the risks associated with the platform's 'vibe-coding' development approach.

Japan's ongoing lower house election has become a testing ground for AI-generated fake videos and deepfakes used as political disinformation tools. Research reveals over half of voters risk mistaking such content for real news, prompting experts to warn this situation represents a serious alarm signal for global democracies.

The rise of AI in mental health has ignited a debate among experts about adding 'friction' to its use. According to Forbes, digital barriers are being proposed to prevent uncontrolled use. Meanwhile, an Ask a Manager survey reveals that highly stressed employees are seeking both traditional and digital solutions.

Google and Peking University researchers have developed a revolutionary system for academic publishing. The PaperBanana framework uses five specialized AI agents to automatically convert method descriptions into publication-ready scientific diagrams, producing expert-level outputs.

Artificial intelligence technologies are evolving beyond simple commands toward more complex and autonomous systems. Production-grade agentic AI systems, equipped with advanced techniques like hybrid retrieval, provenance tracking, and episodic memory, are poised to revolutionize research and analysis processes.

Artificial intelligence technologies are revolutionizing drug development processes by shortening timelines and reducing costs. This transformation is reshaping the financial dynamics of the healthcare sector, from the valuation of biotechnology companies to investor strategies. Advances in statistical modeling and data analysis are paving the way for new drug discoveries.

As artificial intelligence fundamentally transforms business, a critical skills gap in leadership has emerged. A McKinsey study reveals 87% of companies face significant AI skill shortages in management. To address this, IEEE and Rutgers Business School have launched an AI-focused online Mini-MBA program.

OpenAI and Ginkgo Bioworks have developed a platform integrating the GPT-5 AI model into an autonomous laboratory system to optimize cell-free protein synthesis. This collaboration aims to accelerate processes and reduce costs in biotechnology research, though it also highlights limitations in technical infrastructure requirements and autonomous decision-making mechanisms.

Amazon Web Services (AWS) has announced its Nova rubric-based 'LLM Judge' system to evaluate AI model performance within a scientific framework. Available through SageMaker, this system compares different language models, providing developers with objective metrics and critical guidance for model selection.

Popular claims that artificial intelligence has surpassed human intelligence are considered fundamentally flawed by experts. The biological, social, and emotional depth of human intelligence creates a unique realm that current AI systems cannot access. This comparison creates a false sense of competition rather than fostering an understanding of technological progress.

Apple has reportedly shelved its 'Mulberry' project, an AI-powered personal health coach. According to Bloomberg, the company plans to integrate some features into the existing Health app instead. The decision was influenced by a new leadership structure in the health unit under senior vice president Eddy Cue.

AI researchers are working to equip large language models like ChatGPT with an 'inner monologue' capability. This new approach will help AI assess its own confidence and decide when deeper thinking is required, potentially revolutionizing how these systems process and deliver information.

New research from Cyberhaven reveals that the rapid proliferation of corporate AI usage is creating novel data risks that traditional security systems cannot detect. Experts emphasize an urgent need for integrated security platforms in this new era where data fragments move at machine speed.

EPFL researchers have developed a method that solves the 'drift' problem that emerges over time in AI-generated videos. This breakthrough paves the way for AI to produce consistent, long videos without any duration limit.

Apple's secretive 'Mulberry' project, aimed at developing an AI-powered personal health coach, has reportedly been suspended. According to Bloomberg, this decision stems from the company's new operational approach and management restructuring, which prioritizes faster, commercially viable outcomes. Apple is now shifting its focus toward AI features that can be brought to market more quickly.

Apple has reportedly shelved its 'Health+' AI-powered health coaching service, which was designed to analyze users' lifestyles and provide doctor-like recommendations. According to Bloomberg, the company will instead integrate the developed features piecemeal into its existing Health app. The decision is attributed to leadership changes and competitive concerns.

Japanese pharmaceutical giant Takeda aims to overcome data incompatibility—one of AI development's biggest hurdles—through a specialized platform established in Austin, Texas. By building a scalable and compliant infrastructure, the company seeks to revolutionize healthcare by developing AI models trained on real-world data.

Japanese research reveals that AI agents equipped with personality traits and the ability to interrupt during conversations demonstrate significantly improved performance in solving complex problems. This model, which mimics human discussion dynamics, represents a substantial advancement over traditional sequential speech systems.

Apple is grappling with a significant brain drain in its artificial intelligence division, raising questions about its competitive position. The retirement of AI chief John Giannandrea and the departure of approximately a dozen Siri researchers to Meta have created substantial challenges. The company is responding with new appointments and organizational restructuring to manage the crisis.

Commercials aired during Super Bowl LX drew attention with artificial intelligence integration and the growing presence of the healthcare sector. According to CNET's report, this year's advertisements both incorporate and promote AI technology. MMM-Online's compilation shows pharmaceutical giants like Novartis, Boehringer Ingelheim, and Novo Nordisk have increased their advertising investments.

Apple's newly published research paper reveals that design experts can train artificial intelligence models to produce superior user interfaces (UI). Inspired by traditional fashion designers, this methodology helps AI achieve a balance between aesthetics and functionality.

The scientific community is focusing on two groundbreaking developments in breast cancer diagnosis. AI-powered screening systems can detect aggressive tumors at much earlier stages than traditional methods. Meanwhile, a wearable portable ultrasound sensor developed by MIT researchers enables regular home monitoring, potentially ushering in a new era of early detection.