Blog

March 5, 2025

ikayaniaamirshahzad@gmail.com

Weak-to-Strong Alignment via Multi-Agent Contrastive Preference Optimization

[Submitted on 10 Oct 2024 (v1), last revised 2 Mar 2025 (this version, v2)] View a PDF of the paper titled MACPO: Weak-to-Strong Alignment via Multi-Agent Contrastive Preference Optimization, by Yougang Lyu and 6 other authors View PDF HTML (experimental) Abstract:As large language models (LLMs) are rapidly advancing and achieving near-human capabilities on specific tasks, aligning them with human values is becoming more urgent. In scenarios where LLMs outperform humans, we face a weak-to-strong alignment problem where we need to effectively align strong student LLMs through weak supervision generated by weak teachers. Existing alignment methods mainly focus on strong-to-weak alignment

March 5, 2025

ikayaniaamirshahzad@gmail.com

The Challenges and Upsides of Using AI in Scientific Writing

This is a guest post. The views expressed here are the author’s own and do not represent positions of IEEE Spectrum, The Institute or IEEE. Scientific writing is at a pivotal stage, driven by artificial intelligence as a disruptor and enabler. Academics, publishers, and policymakers are attempting to weigh the value of using AI responsibly to enhance productivity versus risking the integrity and purpose of scholarly communication. In this context, the responsible use of the technology in scientific writing pertains to employing AI tools in ways that uphold the integrity, transparency, and ethical standards of scholarly communication. As we collectively

March 4, 2025

ikayaniaamirshahzad@gmail.com

Data Machina #262 – by Carlos

Hoping the AI Agents would show up and help. But after the humans in charge evaporated, the AI Agents never arrived. 3 flights cancelled. 30 hours stranded in Gatwick. No personalised online help, no chatbot assistants, no cash from the ATM, no ccard payments, no flights to escape from hell. Being human is about feeling useless and impotent when 1,000s of machines fail in chain because a little s/w update went haywire. I wonder what will happen when in a few years -inevitably- billions of AI Agents operating out in the wild decide to go on strike. In the meantime,

March 4, 2025

ikayaniaamirshahzad@gmail.com

Google’s AI Co-Scientist: 72-Hour Research Breakthrough

Key Takeaways: Google’s AI co-scientist replicated 10 years of antibiotic resistance research in under 72 hours¹’³. The system combines seven specialized AI agents to mimic human teamwork, from hypothesis generation to fact-checking²’⁴. Ethical safeguards ensure scientists retain control, with AI acting as a collaborator, not a replacement¹’⁴. Early adopters report faster discoveries, including repurposed drugs for liver disease and streamlined cancer research³’⁴. The Co-Scientist That Cracked a 10-Year Puzzle in 48 Hours In February 2025, Professor José Penadés and his team at Imperial College London handed over their unpublished, decade-long research on antibiotic-resistant superbugs to Google’s AI co-scientist. Two days later,

March 4, 2025

ikayaniaamirshahzad@gmail.com

Reddit – Dive into anything

We value your privacy Reddit and its partners use cookies and similar technologies to provide you with a better experience. By accepting all cookies, you agree to our use of cookies to deliver and maintain our services and site, improve the quality of Reddit, personalize Reddit content and advertising, and measure the effectiveness of advertising. By rejecting non-essential cookies, Reddit may still use certain cookies to ensure the proper functionality of our platform. For more information, please see our Cookie Notice and our Privacy Policy. Source link

March 4, 2025

ikayaniaamirshahzad@gmail.com

Reddit – Dive into anything

We value your privacy Reddit and its partners use cookies and similar technologies to provide you with a better experience. By accepting all cookies, you agree to our use of cookies to deliver and maintain our services and site, improve the quality of Reddit, personalize Reddit content and advertising, and measure the effectiveness of advertising. By rejecting non-essential cookies, Reddit may still use certain cookies to ensure the proper functionality of our platform. For more information, please see our Cookie Notice and our Privacy Policy. Source link

March 4, 2025

ikayaniaamirshahzad@gmail.com

Forget ChatGPT — Google Gemini can now see the world with live video and screen sharing

Google‘s AI assistant, Gemini, is set to introduce exciting features to give Android users new ways to interact more intuitively with their devices. Leveraging advanced capabilities, Gemini will soon allow users to ask questions about content on their screens, much like the screen sharing feature currently available in Gemini 2.0 on desktop. In a recent announcement, Google unveiled these Gemini functionalities, which focus on real-time interaction and on-screen inquiries. These features are part of Google’s Project Astra. New functionalities (Image credit: Google Gemini) The screen-sharing function allows users to share their screens with Gemini and ask questions based on displayed

March 4, 2025

ikayaniaamirshahzad@gmail.com

[2412.17762] The Superposition of Diffusion Models Using the Itô Density Estimator

[Submitted on 23 Dec 2024 (v1), last revised 28 Feb 2025 (this version, v2)] View a PDF of the paper titled The Superposition of Diffusion Models Using the It\^o Density Estimator, by Marta Skreta and Lazar Atanackovic and Avishek Joey Bose and Alexander Tong and Kirill Neklyudov View PDF HTML (experimental) Abstract:The Cambrian explosion of easily accessible pre-trained diffusion models suggests a demand for methods that combine multiple different pre-trained diffusion models without incurring the significant computational burden of re-training a larger combined model. In this paper, we cast the problem of combining multiple pre-trained diffusion models at the generation

March 4, 2025

ikayaniaamirshahzad@gmail.com

Nothing Shows Off Its Phone (3A) Line

Image: Nothing Like last year, the London-based independent smartphone company Nothing announced a new model during Mobile World Congress. The Phone (3A) line continues Nothing’s trend of distinct visual design with modernized cameras and an artificial intelligence feature: Essential Space. Two models make up the Phone (3A) line: (3A) and (3A) Pro. The basic (3A) model costs $379; it can be preordered starting March 4, shipping March 11. The (3A) Pro model costs $459; it can be ordered starting March 11, shipping March 25. UK-based Nothing phone has limited U.S. availability To get a Nothing phone in the U.S., you’ll

March 4, 2025

ikayaniaamirshahzad@gmail.com

Why Post-Training Matters Now: From SFT to RFT

In today’s competitive AI landscape, customization of foundation models has become essential for organizations seeking to create differentiated value. As using the same models as competitors leads to commoditization, post-training techniques have emerged as critical tools that allow enterprises to tailor models to their specific needs without incurring the prohibitive costs of building models from scratch. Among these techniques, Supervised Fine-Tuning (SFT) and Reinforcement Fine-Tuning (RFT)  represent two distinct approaches with unique strengths and applications. The economics of model customization have shifted dramatically in favor of post-training methods, with efficiency gains increasingly derived from strategic adaptations rather than developing entirely