AI Research
Realism Control One-step Diffusion for Real-World Image Super-Resolution

arXiv:2509.10122v1 Announce Type: cross
Abstract: Pre-trained diffusion models have shown great potential in real-world image super-resolution (Real-ISR) tasks by enabling high-resolution reconstructions. While one-step diffusion (OSD) methods significantly improve efficiency compared to traditional multi-step approaches, they still have limitations in balancing fidelity and realism across diverse scenarios. Since the OSDs for SR are usually trained or distilled by a single timestep, they lack flexible control mechanisms to adaptively prioritize these competing objectives, which are inherently manageable in multi-step methods through adjusting sampling steps. To address this challenge, we propose a Realism Controlled One-step Diffusion (RCOD) framework for Real-ISR. RCOD provides a latent domain grouping strategy that enables explicit control over fidelity-realism trade-offs during the noise prediction phase with minimal training paradigm modifications and original training data. A degradation-aware sampling strategy is also introduced to align distillation regularization with the grouping strategy and enhance the controlling of trade-offs. Moreover, a visual prompt injection module is used to replace conventional text prompts with degradation-aware visual tokens, enhancing both restoration accuracy and semantic consistency. Our method achieves superior fidelity and perceptual quality while maintaining computational efficiency. Extensive experiments demonstrate that RCOD outperforms state-of-the-art OSD methods in both quantitative metrics and visual qualities, with flexible realism control capabilities in the inference stage. The code will be released.
Source link
AI Research
Back to School – With Help From AI – Terms of Service with Clare Duffy

Kirk suspect reportedly confesses, Tesla stock, ‘tooth-in-eye’ surgery & more
5 Things
Listen to
CNN 5 Things
Mon, Sep 15
podcast
New technologies like artificial intelligence, facial recognition and social media algorithms are changing our world so fast that it can be hard to keep up. This cutting-edge tech often inspires overblown hype — and fear. That’s where we come in. Each week, CNN Tech Writer Clare Duffy will break down how these technologies work and what they’ll mean for your life in terms that don’t require an engineering degree to understand. And we’ll empower you to start experimenting with these tools, without getting played by them.
Back to School – With Help From AI Terms of Service with Clare Duffy Sep 16, 2025
Kids are heading back to school. One thing students, teachers and parents can expect to encounter this year is artificial intelligence, which has raised all kinds of questions, both positive and negative. So, how can you make sure your student is navigating AI safely and successfully? Dr. Kathleen Torregrossa has been an educator for 37 years in Cranston, Rhode Island. She explains how teachers are using AI in the classroom, and what families need to know about its impact on learning. – This episode includes a reference to suicide. Help is available if you or someone you know is struggling with suicidal thoughts or mental health matters. In the US: Call or text 988, the Suicide & Crisis Lifeline. Globally: The International Association for Suicide Prevention and Befrienders Worldwide have contact information for crisis centers.
AI Research
Lewis Honors College introduces ‘Ideas that Matter’ program series

LEXINGTON, Ky. (Sept. 16, 2025) — This fall, the Lewis Honors College (LHC) launches its “Ideas that Matter” series, a program connecting students with leading scholars, innovators and changemakers on issues shaping today’s world — from free speech and artificial intelligence to nonprofit innovation.
LHC Director of College Life Libby Hannon, who initiated the series, said the goal is to spark lively dialogue.
“The ‘Ideas that Matter’ discussions combine intellectually engaging questions with interactive conversations and allow our students to speak with some of the most forward-thinking scholars, changemakers and entrepreneurs from Lexington and beyond,” Hannon said.
The series begins Sept. 18 with University Research Professor Neal Hutchens, Ph.D., who will explore the historical and legal background of free speech and academic freedom in campus life. His talk, 5-6 p.m. in the Lewis Scholars Lounge, will conclude with an interactive Q&A.
“I’m especially looking forward to the conversation part of the evening, where we engage in and model the kind of vibrant back-and-forth that is crucial to maintaining systems of free speech and academic freedom,” Hutchens said.
On Oct. 6, Lewis Lecturer Sherelle Roberts, Ph.D., will moderate a panel of experts on artificial intelligence as they discuss “The Future of Earth and AI,” including the current and potential impacts of artificial intelligence on the future of work, the economy and the environment.
“Artificial Intelligence is quickly becoming a part of our everyday lives. Some even believe AI will transform our world as dramatically as the Industrial Revolution,” Roberts said. “This event will get our students thinking critically about our possible AI-driven future, while also having some fun.”
The event will begin at 5:30 p.m. with movie snacks and will transition into the panel discussion at 6 p.m., featuring faculty and staff from a variety of disciplines. The movie, an animated film that conceptualizes our AI-powered future, will begin at 7 p.m.
The final event of the semester on Nov. 11, will spotlight local nonprofit Operation Secret Santa (OSS), 5-6 p.m. in the Lewis Scholars Lounge. Founder Katie Keys and honors program alum Lucy Jett Waterbury will share the story of OSS’s creation in 2016 and its growing impact on the community.
“Operation Secret Santa is built on the belief that no child should face barriers to feeling loved and celebrated,” said Keys. “We meet families where they are, right at their doorsteps, bringing not only gifts and food, but the reminder that their village sees them and cares.”
“From (Katie’s) big heart, she has built a big, yet lean and efficient, nonprofit that has one very simple goal, to bring joy to Kentucky kids at Christmas time,” Waterbury said.
Through this series, LHC offers students a chance to engage with pressing issues, broaden their perspectives and learn directly from those making a difference.
AI Research
Ethereum Foundation Bets Big on AI Agents with New Research Team

TLDR
- Ethereum Foundation launches new dAI Team led by research scientist Davide Crapis to connect blockchain and AI economies
- Team focuses on enabling AI agents to make payments and coordinate without intermediaries on Ethereum
- Group continues work on ERC-8004 standard for proving AI agent identity and trust
- Initiative aims to make Ethereum the settlement layer for autonomous machine transactions
- Foundation hiring AI researcher and project manager to staff the new specialized unit
The Ethereum Foundation has formed a specialized artificial intelligence research team to position Ethereum as the foundation for autonomous machine transactions. Research scientist Davide Crapis announced the new dAI Team on Monday, outlining plans to merge blockchain technology with AI systems.
The team will pursue two main goals according to Crapis. First, enabling AI agents to conduct payments and coordinate activities without human intermediaries. Second, building a decentralized AI infrastructure that reduces dependence on major technology companies.
We’re starting a new AI Team at the Ethereum Foundation (the dAI Team).
Our mission: make Ethereum the preferred settlement and coordination layer for AIs and the machine economy.The team will focus on two main areas:
– AI Economy on Ethereum = giving AI agents and robots ways… pic.twitter.com/9sWVS4dp0K— Davide Crapis (@DavideCrapis) September 15, 2025
Crapis leads the new unit and will connect its work with the Foundation’s protocol development group and ecosystem support division. The team has begun hiring for an AI researcher position and a project manager role to drive coordination efforts.
The dAI Team builds on existing work around ERC-8004, a proposed Ethereum standard co-authored by Crapis. This standard aims to establish identity and reputation systems for autonomous AI agents. The protocol would allow these agents to prove their trustworthiness and coordinate activities without centralized oversight.
AI Agent Infrastructure Development
The Ethereum Foundation sees growing demand for settlement systems as AI agents begin conducting more transactions. Crapis stated that intelligent agents need neutral infrastructure for handling value transfers and reputation management. Ethereum’s censorship resistance and verifiability make it suitable for these functions.
Current blockchain activity supports this vision of expanded use cases. CryptoQuant data shows Ethereum processed 12 million daily smart contract calls on Thursday. The analytics firm noted that network activity remains in expansion mode with record transaction volumes and active addresses.
AI agents operate as programs that make decisions with minimal human supervision. They can execute transactions and perform tasks on behalf of their programmers. Blockchains with programmable features like smart contracts provide suitable environments for these autonomous systems.
The Foundation restructured in 2025 to handle Ethereum’s growth through specialized units. The dAI Team represents part of this shift toward addressing emerging technologies. Previous focus areas included layer-2 scaling solutions and zero-knowledge proof development.
Decentralized AI Stack Goals
Multiple blockchain projects are working to integrate AI and distributed ledger technology. Matchain launched a decentralized AI blockchain in 2024. KiteAI announced an AI-driven blockchain in the Avalanche ecosystem in February 2025.
The Ethereum Foundation’s approach differs by focusing on standards and infrastructure rather than creating new blockchains. The dAI Team will support public goods and projects that combine AI with existing Ethereum capabilities.
Crapis emphasized the mutual benefits of linking AI and Ethereum. He stated that Ethereum makes AI more trustworthy while AI makes Ethereum more useful. This relationship could expand as more autonomous agents require blockchain services.
The team operates under Ethereum’s decentralized acceleration philosophy. This approach prioritizes open and verifiable AI development while maintaining human oversight of intelligent systems. The Foundation aims to prevent AI infrastructure lock-in by major technology companies.
Industry experts see potential for AI agents and blockchain technology to reshape digital commerce. The combination could enable new forms of autonomous economic activity without traditional intermediaries.
The Ethereum Foundation has begun publishing resources for the new team according to Crapis. He stated the Foundation will work with urgency to connect AI developers with the Ethereum ecosystem and accelerate research between the two fields.
-
Business3 weeks ago
The Guardian view on Trump and the Fed: independence is no substitute for accountability | Editorial
-
Tools & Platforms1 month ago
Building Trust in Military AI Starts with Opening the Black Box – War on the Rocks
-
Ethics & Policy2 months ago
SDAIA Supports Saudi Arabia’s Leadership in Shaping Global AI Ethics, Policy, and Research – وكالة الأنباء السعودية
-
Events & Conferences4 months ago
Journey to 1000 models: Scaling Instagram’s recommendation system
-
Jobs & Careers3 months ago
Mumbai-based Perplexity Alternative Has 60k+ Users Without Funding
-
Podcasts & Talks2 months ago
Happy 4th of July! 🎆 Made with Veo 3 in Gemini
-
Education3 months ago
VEX Robotics launches AI-powered classroom robotics system
-
Education2 months ago
Macron says UK and France have duty to tackle illegal migration ‘with humanity, solidarity and firmness’ – UK politics live | Politics
-
Podcasts & Talks2 months ago
OpenAI 🤝 @teamganassi
-
Funding & Business3 months ago
Kayak and Expedia race to build AI travel agents that turn social posts into itineraries