Tools & Platforms

Kennesaw State Researcher Pioneers New Frontiers for AI Beyond Cloud

Published

3 days ago

September 10, 2025

Artificial Intelligence (AI) has become synonymous with complex systems, mega data centers, and sophisticated algorithms. However, a groundbreaking initiative led by Bobin Deng, an assistant professor at Kennesaw State University’s College of Computing and Software Engineering, aims to democratize access to AI. With recent support from the National Science Foundation (NSF), Deng is working on pushing the boundaries of AI applications beyond the confines of high-powered servers and into personal devices, where it can have a profound impact on everyday lives. This shift is not just about convenience; it is about fundamentally changing how AI interacts with individuals and small systems.

Integrating AI into personal devices is more than a mere upgrade; it represents a pivotal moment in technology. Current AI systems predominantly operate on robust, expensive servers, often requiring a stable internet connection, which limits their accessibility and usability. Deng proposes a revolutionary approach, allowing AI algorithms to function offline on devices such as smartphones, drones, and even industrial sensors. By doing so, he envisions a future where advanced technology is not just reserved for large corporations or research institutions but can be harnessed by anyone, anywhere.

The core concept of Deng’s research revolves around a technique known as activation sparsity. This ingenious method exploits the fact that in any given AI model, only a fraction of neurons are actively engaged during computation. Instead of overwhelming devices with demands for all data to be loaded simultaneously, this approach allows the system to anticipate which data points will likely be required and pre-load only those necessary components. This not only conserves memory resources but also accelerates processing speeds while significantly lowering energy consumption—an essential factor in prolonging the lifecycle of portable devices.

In essence, the current methodology employed to reduce the size and complexity of AI models typically involves either diminishing the precision of data or eliminating less significant parameters. However, the innovative activation sparsity technique proposed by Deng diverges from these traditional methods. By predicting which values will activate during the course of a model’s operation, he can combine this strategy with existing methods like pruning and quantization. This multi-faceted approach ultimately leads to an unprecedented level of efficiency in AI systems that could redefine how developers conceptualize and deploy their software solutions.

Deng’s research team is engaging in empirical studies to test tiny machine-learning models as predictive support for larger systems. This interplay between small-scale predictive models and more extensive systems may bridge gaps that previously existed between advanced AI applications and user-friendly technologies. The implications of this work stretch beyond the technical realm; they highlight a shift towards embedding AI capabilities into the very devices we use daily, thus streamlining processes and enhancing user experience.

A major aspect of Deng’s initiative includes the development of an open-source simulator, which is designed to allow students and fellow researchers to interact with and refine this pioneering technology. By fostering an environment of collaboration and shared learning, the simulator could accelerate innovation in the field of AI, providing a robust platform for testing and iteration that goes beyond the walls of Kennesaw State University.

The significance of this NSF-funded project cannot be overstated, as it is positioned to impact educational programs, particularly at Kennesaw State, which recently launched a Master of Science in Artificial Intelligence program. The potential applications of Deng’s research extend beyond academia into numerous industries, with prospects ranging from monitoring factory robotics to addressing predictive maintenance in various sectors. This fusion of academic inquiry and practical application stands to enhance the way industries operate, making smarter and more effective use of AI strategies.

Within the supportive ecosystem at Kennesaw State, Deng has noted the substantial assistance he has received, which has been conducive to his research and its administrative aspects. The university has provided him the necessary resources and a conducive environment to pursue such ambitious projects. This collaboration highlights the crucial role that educational institutions play in nurturing groundbreaking research and fostering innovation that advances technology.

As AI technology continues to evolve, forward-thinking researchers like Deng are crucial in steering its development towards responsible and beneficial uses. By focusing on accessibility and efficiency, his work promises to remove barriers to AI engagement and make advanced computational tools available to a broader audience. Institutions such as Kennesaw State are paving the way for a new generation of innovations that can bridge the gap between high-tech and everyday uses.

The overarching vision espoused by Deng and his team underscores a key aspect of future technological development: that the most powerful tools should not only be in the hands of the few but should empower individuals and small enterprises. This vision aligns with Kennesaw State’s mission to prepare students for an increasingly technology-driven future while simultaneously addressing real-world challenges in diverse domains.

As we stand on the brink of a new age of technology, where artificial intelligence becomes an integral part of our daily lives, the work being done by Kennesaw State University’s Bobin Deng exemplifies the innovative spirit and forward-thinking mindset that will drive this evolution. By making AI more accessible and efficient, we can anticipate a world where the possibilities of technology are continually expanded, fueling significant advancements across all sectors of society.

Through research initiatives like Deng’s, the landscape of artificial intelligence is not just being explored; it is being redefined. As these advancements unfold, they promise to enhance the global understanding of AI, fostering greater acceptance and use in a multitude of contexts, ultimately leading towards a more connected and intelligent world.

The convergence of advanced AI techniques with personal and industrial devices could soon mark a transformative chapter in technology, fostering a scenario where cutting-edge capabilities are available to everyone and facilitating a future filled with potential innovations and breakthroughs that benefit society at large. Such work underlines the importance of continuing investment in scientific research and education, essential components for nurturing the next generation of leaders in technology and engineering.

Subject of Research: Accessibility of Artificial Intelligence through Activation Sparsity.
Article Title: Revolutionizing AI: Making Artificial Intelligence Accessible Beyond the Cloud.
News Publication Date: October 2023.
Web References: Kennesaw State University, NSF.
References: National Science Foundation Grant Documents.
Image Credits: Darnell Wilburn / Kennesaw State University.

Keywords

Artificial Intelligence, Activation Sparsity, Kennesaw State University, National Science Foundation, Sustainable Smart Systems, Machine Learning.

Tags: activation sparsity techniqueAI accessibility in personal devicesAI applications beyond cloudBobin Deng AI initiativedemocratizing artificial intelligencefuture of decentralized AIKennesaw State University researchNational Science Foundation supportoffline AI algorithmspersonal device AI integrationsmall systems AI interactiontransforming everyday technology

Source link

Tools & Platforms

The evolving role of HR in the AI-first enterprise – People Matters India

Published

44 minutes ago

September 14, 2025

People Matters Pte. Ltd.

The evolving role of HR in the AI-first enterprise People Matters India

Source link

Tools & Platforms

The investment of the year and AI like fire

Published

1 hour ago

September 14, 2025

Bart

As we all know, there’s a lot of talk about AI lately. Some of it is pretty hyped-up (“it will be a disaster!” vs. “heaven is just around the corner!”), but IO+ is here to select the pearls from the rubbish. Pearl of the week is the interview Carlo van de Weijer had with the Possible Futures Podcast. Van de Weijer is basically optimistic, but with a realistic perspective. “Technology has always been the main thing that saves our future,” he argues. “Every breakthrough created new problems, but also the tools to solve them.”

AI won’t change the basics of human life, Van de Weijer expects. “It won’t rewrite the rhythms of human existence. It just makes them smoother.” Still, he warns against complacency. “AI is like fire. Fire gave us warmth and cooking, but it could also burn down your house. The difference is that AI moves much faster and can continually improve itself. That’s why we need regulation.” Van de Weijer also discusses jobs lost & found, increased productivity, and its impact on education. More here.

ASML

The news leaked on Sunday afternoon and was made official on Wednesday: ASML has acquired more than 10% of the French AI company Mistral for €1.3 billion. Hardware is embracing software; iron into AI. We were on top of it with the first news, the nuances from ASML itself, and an analysis of this rather surprising move. We highly recommend the analysis, in which we present the eight most important arguments to show that this seemingly illogical step is actually very logical. “Strategically brilliant,” even, as we heard from ASML itself.

Officially, the goal of this “strategic partnership” is to explore the use of AI models in ASML’s product portfolio and in research, development, and operations. Simply put, it will enhance ASML’s large machines, making them better and more reliable, as well as the research that will lead to the successors of these mega-devices. But that’s not all, because… well, read for yourself 🙂

Gerard & Anton Community

Don’t want to miss anything about the startup ecosystem of the high-tech manufacturing industry? Then subscribe to the newsletters of the Gerard & Anton community. Every month, you’ll receive an overview of the highlights – check out yesterday’s edition here.

The Gerard & Anton monthly newsletter is a great way to get a flavor of one of the most active startup ecosystems in Europe. It follows raw ideas (via Drinks, Pitches & Demos), bold visions (via Brabant Bits & Bytes), early successes (via the G&A Awards), lifetime achievements (via the Piek Awards), and all other reasons to celebrate (via the Founders Dinner, a.o.). In addition to all that, Gerardanton.com/news provides your daily update on relevant developments in the ecosystem.

Watt Matters in AI

We are approaching November 26, the day we will be holding our Watt Matters in AI conference. Everything about the exorbitant energy consumption of AI systems, but above all: what solutions are there to prevent this from getting completely out of hand? We have now secured seven very interesting speakers: a guarantee that by the end of the day you will be completely up to speed on this topic. Tickets are available via Watt Matters in AI.

A podcast every day

We’ll keep saying it: every morning at 6:30 a.m. (on weekdays), a new podcast is waiting for you. In it, our AI colleagues Oliver and Shelby discuss the two most interesting topics of the day. This makes the IO+ Daily the ideal way to fill your head with some optimistic news from the world of innovation and technology.

Our other newsletters

Thank you for reading this newsletter. But we have more to offer. Signing up is very easy (just click on the name of the newsletter):

A selection of other highlights from this week:

Enjoy your Sunday and don’t forget that a new episode of IO+ Daily will be waiting for you tomorrow morning at 6:30 a.m.

Source link

Tools & Platforms

Nano Banana AI daily limit: How many images can you generate with Google Gemini; full guide to free, Pro and Ultra access |

Published

1 hour ago

September 14, 2025

TOI Tech Desk

Google’s Gemini Nano Banana has taken the internet by storm, becoming one of the most widely discussed AI tools in 2025. Unlike traditional image generators, Nano Banana specialises in turning ordinary photos into 3D models with an impressive degree of realism. Users can create lifelike figurines, avatars, and even cinematic-style effects with just a short text prompt.The model stands out for its speed and detail retention, making it faster than most competitors while ensuring generated models don’t lose essential features. This mix of efficiency, creative flexibility, and viral appeal has driven its rapid adoption across social media.

Google updated Gemini Nano Banana Image limits; here’s why

With demand soaring, Google recently changed its policy on image generation limits. Previously, the company provided clear daily numbers—100 images per day for free users and up to 1,000 for Pro and Ultra subscribers. But as usage grew, these fixed limits were quietly removed.Instead, Google now refers to access tiers as “basic access” for free accounts and “highest access” for Pro and Ultra plans. This shift allows the company to dynamically adjust user limits depending on server capacity, subscription status, and global demand.This update indicates that free users may face more restrictions during peak hours, while paying customers will continue to enjoy smoother, prioritized access.

Free image generation on Gemini Nano Banana

Originally, free users were allowed 100 image generations per day, making Gemini Nano Banana one of the most generous free AI platforms available. For comparison, ChatGPT limits free users to just two image generations daily.With the new update, the support page now simply lists “basic access” for free accounts. While Google no longer guarantees a fixed number of images, users can still expect to generate far more than most competing free AI tools. However, during high traffic times, Google may temporarily throttle free usage to keep the platform stable.

Gemini Pro and Ultra users: What “highest access” really means

For paid subscribers, Google now describes image generation as offering “highest access.” Previously, Pro and Ultra users enjoyed a cap of 1,000 images per day, but this number is no longer publicly displayed.This change gives Google more flexibility. Instead of promising a strict daily quota, the company can adjust Pro and Ultra limits to ensure they always receive priority processing speed, minimal wait times, and higher usage availability compared to free users.For professionals, digital artists, or content creators who rely on Nano Banana daily, Pro and Ultra remain the best options for consistent, large-scale image generation.

How Google keeps AI image tools running smoothly

Google’s decision to replace clear numbers with tier-based access categories reflects a broader strategy. As AI image generation tools scale up, server demand and computing resources fluctuate constantly. By setting flexible tiers, Google can manage heavy traffic more effectively while ensuring subscribers remain satisfied.For free users, this means access may feel unpredictable at times. But for Google, it ensures the service doesn’t crash under the weight of viral trends, such as the current Nano Banana 3D figurine craze.

Gemini Nano Banana vs. competing AI tools

Despite the new policy, Gemini Nano Banana still holds an edge over competitors. Free users can generate far more content than with ChatGPT or MidJourney’s trial options, making it a more attractive entry point for casual creators.Moreover, Nano Banana’s advanced 3D rendering capabilities and ability to process natural language prompts into complex effects set it apart in terms of creative potential. Instead of manually adjusting settings, users can simply describe their vision and let the AI handle the details.

Nano Banana AI: Steps to generate 3D models with Gemini app

Creating images and 3D models with Gemini Nano Banana is simple and beginner-friendly:

Open the Gemini app and navigate to the Nano Banana option.
Upload an image you want transformed into a 3D model.
Enter a pNano Banana AI: ChatGPT vs Qwen vs Grok vs Gemini; the top alternatives to try in 2025rompt describing the effects or style you want. You can use Google’s official examples or craft your own.
Generate the model, which usually renders in seconds and can be customized further.

Google’s sample prompts show how detailed the outputs can be. For example, users can request a figurine with specific packaging artwork, desk placement, and environmental details—all generated automatically.Also Read | Nano Banana AI: ChatGPT vs Qwen vs Grok vs Gemini; the top alternatives to try in 2025

Source link