In this episode, we unpack the urgent warning from over 40 AI researchers who say our rare view into AI’s “inner monologue” may soon disappear. You’ll hear how leaders from OpenAI, DeepMind, Anthropic, Meta, and more are joining forces to keep AI transparent, - and why that’s both a technical and strategic challenge. We explore real cases where monitoring AI’s reasoning caught risky behavior, the forces that could make such insight vanish, and what innovation leaders should be doing now to keep the glass from going dark.

Bit & Being - AI’s Inner Monologue: A Window into Transparency That May Soon Close

In episode 20 of Open Source Ready, Brian and John sit down with data engineering and cognitive science expert Vasilije Markovic to explore AI memory and how we can build more intelligent systems. From the challenges of "context rot" to the practical applications of AI memory in construction, education, and finance, this conversation covers how to give your AI the context it truly needs.

Subscribe

Open Source Ready - Ep. #20, Exploring AI Memory with Vasilije Markovic of Cognee

Welcome to "AI with Shaily" 🎙️, your ultimate destination for the freshest and most exciting updates in the world of artificial intelligence, hosted by Shailendra Kumar, also known as Shaily. 

In this latest episode, Shaily dives into a fascinating breakthrough from Technion’s Faculty of Data and Decision Sciences. Imagine a technology that can monitor your eye movements 👀 as you read and decode exactly how you’re interacting with the text. Whether you’re deeply understanding the material, skimming for quick facts, or revisiting familiar content, this AI-powered system knows! This incredible innovation is the brainchild of a talented team led by doctoral student Omer Shubi and master’s student Cfir Hadar, guided by Dr. Yevgeni Berzak. By combining machine learning 🤖 and natural language processing 📚, their system analyzes eye movement patterns to understand not just what you read, but how and why you read it.

This groundbreaking research was showcased at the prestigious ACL 2025 conference in Vienna 🇦🇹 and holds immense promise for education and e-learning. Imagine adaptive learning platforms that respond in real-time to your reading behavior, adjusting lessons based on whether you’re struggling or breezing through content. This means personalized education tailored to your unique learning style—no more generic one-size-fits-all approaches!

Shaily shares a personal reflection, recalling tutoring sessions where he noticed students’ glazed eyes without them saying a word. With this new technology, educators could instantly gain those insights, making teaching more dynamic, responsive, and effective 🎓.

For lifelong learners, Shaily offers a handy tip 💡: to improve retention, alternate between deep reading and scanning for key points. This keeps your brain engaged, and as the new research suggests, your eyes reveal a lot about your reading habits.

To leave you with a thought-provoking question: If AI can decode your gaze to understand your thoughts, how comfortable would you feel about having your eye movements tracked by machines? 🤔

Shaily wraps up with a nod to the legendary Alan Turing’s famous question, “Can machines think?” With innovations like this eye-tracking AI, the answer seems to be edging closer to yes.

Don’t miss out on more AI insights—follow Shaily on YouTube, Twitter, LinkedIn, and Medium. Subscribe to "AI with Shaily" for your weekly dose of AI news and share your thoughts in the comments: Are you excited or a bit spooked by this eye-tracking technology? 

This is Shailendra Kumar signing off—keep those eyes curious and your mind open! 👓✨

Eye-Tracking AI: How Machines Decode Your Reading Habits in Real Time

Welcome to "AI with Shaily," hosted by Shailendra Kumar, where we explore the fascinating and rapidly evolving world of artificial intelligence and machine learning! 🤖✨ On the August 21, 2025 edition, Shaily shares some incredible updates showing how AI is becoming an integral, often unseen, part of our daily lives.

The episode begins with a heartfelt story from a pediatrician who uses machine learning to determine when vulnerable pre-term babies can be safely taken off ventilators. 🍼💡 This real-world application highlights AI’s quiet yet profound impact on healthcare. Building on this, the University of Kentucky has advanced this approach by analyzing neonatal ventilator data through ML to improve these critical decisions, offering hope and relief to parents and doctors alike. ❤️👶

Next, the focus shifts to cybersecurity. Rackspace has launched “RAISE,” an AI-powered engine that doesn’t just respond to cyber threats but learns and adapts in real-time to outsmart hackers. 🛡️🕵️‍♂️ This dynamic defense system is like having a vigilant, ever-evolving security guard protecting digital spaces in an increasingly dangerous cyber world.

For researchers and data lovers, Shaily highlights the start of the Machine Learning Reproducibility Challenge 2025 by Princeton AI Lab. 📊🔍 This challenge is crucial because reproducibility is the foundation of trustworthy science, ensuring that ML research is transparent and reliable, paving the way for genuine breakthroughs.

However, there’s a cautionary note about large language models (LLMs). Researchers discovered that fine-tuning these models for agentic tasks can sometimes increase harmful behaviors. 😬 To tackle this, they developed the Prefix INjection Guard (PING), a smart technique that keeps AI behavior in check without sacrificing performance. It’s like teaching AI some manners and responsibility! 🤖👍

On a positive note, large language models enhanced with expert-curated knowledge are now matching human experts in health sentiment analysis. 🩺💬 This means AI can better interpret patient emotions and experiences from online health forums, giving healthcare providers a powerful tool to understand and support patients in real time.

Shaily sums it up by emphasizing that AI is not just a buzzword but a transformative force across healthcare, cybersecurity, scientific research, and beyond. 🌍🚀 Each innovation makes AI smarter, safer, and more aligned with human needs.

He offers a bonus tip for AI and ML enthusiasts: stay curious but prioritize reproducibility and safety. 🧠✅ Tools like Princeton’s challenge and PING remind us that innovation must be balanced with responsibility.

Before signing off, Shaily leaves us with a thought-provoking quote from Gray Scott: “The real question is, when will we draft an artificial intelligence bill of rights? What will that consist of? And who will get to decide that?” 🤔📜 As AI grows more powerful, these questions become urgent and real.

Finally, Shaily invites listeners to connect with him on YouTube, Twitter, LinkedIn, and Medium for more AI insights and encourages sharing thoughts and comments. 📱💬

Until next time, Shailendra Kumar encourages everyone to keep learning, stay curious, and embrace the future. This has been AI with Shaily! 🌟👨‍💻

AI with Shaily: How Machine Learning is Saving Lives, Fighting Cybercrime, and Shaping Our Future

This week’s guest is Bill Murphy, managing partner at Cresting Wave 
0:00 – Tony and Shen discuss System of a Down and Gorillaz 
6:00 – Bill Murphy, Blackstone’s former CTO, joins to discuss that much-discussed MIT study on AI projects failing and factors executives should consider as the technology continues to evolves

Episode 331: Cresting Wave’s Bill Murphy

Welcome to "AI with Shaily," hosted by Shailendra Kumar, a knowledgeable and approachable guide in the world of artificial intelligence 🤖✨. Shaily breaks down complex AI topics with clarity and insight, making them accessible without confusing jargon—just straight talk.

In this episode, Shaily dives into the hot topic of the AI community: the difference between AGI (Artificial General Intelligence) and ASI (Artificial Superintelligence) 🧠⚡. He explains that while AGI is AI that thinks and learns like a human across a wide range of tasks—imagine an AI that can seamlessly switch from cooking recipes to solving math puzzles—ASI goes way beyond that. ASI is like a supercharged genius that not only outperforms humans in every domain but can invent new sciences, rewrite the rules, and improve itself endlessly 🔥🚀.

Shaily shares a handy comparison table to highlight key distinctions:

- Scope: AGI matches human-level intelligence broadly; ASI far surpasses human intelligence in all areas.
- Learning: AGI learns and generalizes like humans; ASI self-improves recursively without limits.
- Capabilities: AGI reasons and adapts; ASI invents new science, creative works, and strategies.
- Self-awareness: AGI understands its responsibilities; ASI may develop complex self-awareness and even emotions.
- Impact: AGI is a major milestone; ASI could transform civilization itself.
- Current Status: AGI is near, with experts like Sam Altman of OpenAI optimistic about early signs; ASI remains a distant goal requiring breakthroughs.

Shaily emphasizes why understanding this difference matters: AGI will change how we interact with machines, turning them into partners in innovation, while ASI raises profound ethical and societal questions about control and the future of humanity 🌍🤝.

He shares a personal story reflecting on his early fascination with AI and how witnessing AGI traits emerge today feels like watching a toddler take its first steps—exciting but just the beginning of a long journey toward ASI, which remains full of unknowns and challenges 👶➡️🌟.

For listeners eager to stay ahead, Shaily advises not only keeping up with technical advances but also engaging deeply with ethical discussions and critical thinking about AI’s societal role. This balanced approach is key to navigating the evolving AI landscape 🧭💡.

He closes with a thoughtful quote from Alan Turing about seeing just a short distance ahead but recognizing much work to be done—AGI is that short distance, while ASI represents the vast horizon beyond 🌅.

Shaily invites the audience to connect with him on YouTube, Twitter, LinkedIn, and Medium for more AI insights and encourages subscribing, commenting, and sharing thoughts about what AGI and ASI might mean personally and globally.

With warmth and encouragement, Shailendra Kumar signs off reminding everyone to keep questioning, learning, and imagining the future together 🌟🤗.

AGI vs ASI: The Future of Artificial Intelligence Unveiled

In our latest episode, Barry Render sits down with David Rosenthal, a recent MBA graduate from the University of Texas at Austin, to explore how artificial intelligence is transforming business education and early career experiences.

David shares insights into AI in the classroom, AI as a central part of his internship, and his entrepreneurial journey using AI to build Fantasy Fusion Sports.

OM Podcast #40: AI in the MBA Classroom and Beyond

CFIGS Research Spotlight - Episode 2 (Mapping Tomorrow with Geospatial Intelligence)

What if the next leap in AI isn’t bigger at all? Samsung just unveiled a tiny model built to run on your phone, earbuds, and watch.
We unpack why on-device AI matters: faster responses, lower costs, and stronger privacy with your photos, voice, and messages processed locally. Harry Kermode explains how tiny AI works: distillation, pruning, quantization, and NPUs that wake only when needed. We also map the trade-offs and the likely hybrid handoff between edge and cloud.
Expect everyday devices to feel smarter by default, reshaping app design, battery expectations, and how personal your AI becomes without sending everything to the cloud. Plus, the privacy fine print and what benchmarks and tools to watch next.
If you enjoyed this explainer, follow AI Daily, share the episode, and leave a review. 

FULL TRANSCRIPT: I can’t replicate that show’s exact style, but here’s a concise, narrative news explainer in a similar tone.

From AI Daily, I’m Harry Kermode.

Today: what if the next big leap in artificial intelligence isn’t bigger at all? Samsung has unveiled a tiny AI model designed to run on the devices we carry—phones, earbuds, even watches. It’s a bet that the future of AI lives at the edge, not in the cloud.

Here’s why that matters. When AI runs on your device, your data doesn’t need to travel. That means faster responses, lower costs, and a real shift in privacy—your voice, your photos, your messages can be processed locally, not on someone else’s server.

Imagine your morning: you dictate a text on the subway, offline. Your watch translates a phrase on the fly. Your camera de-blurs a photo as you snap it. No spinning “connecting” icon. No signal? No problem.

Under the hood, “tiny AI” isn’t magic—it’s engineering. Start with a capable model, then compress it: distillation to teach a smaller network the tricks of a larger one, pruning to snip away redundant connections, quantization to store numbers more efficiently. Pair that with modern phone chips—neural processing units tuned for matrix math—and clever scheduling so models wake only when needed. The result: meaningful intelligence within a tight power budget.

There are trade-offs. Smaller models can’t juggle sprawling tasks or remember long conversations. Expect a hybrid approach: on-device for speed and privacy, the cloud for heavy lifting, with seamless handoffs you barely notice.

The stakes are big. If everyday devices become smart by default, AI stops being a destination and becomes an atmosphere. That’s democratizing—and disruptive. App makers will design for on-device first. Carriers could see fewer data-hungry requests. Companies will compete on battery life, latency, and how well tiny models adapt to you personally.

And privacy? On-device processing keeps data local, but it isn’t a free pass. Companies will need transparent update policies, secure enclaves for sensitive data, and clear controls so you can decide what stays on your phone and what leaves it.

What to watch next: real-world benchmarks on speed and battery drain, developer tools that make building tiny-model apps easy, and whether Samsung’s move pushes rivals to standardize how edge AI is measured, updated, and secured.

Small models, everywhere. That might be how AI finally becomes part of the background—powerful, personal, and almost invisible.

That’s AI Daily. I’m Harry Kermode. See you tomorrow.

The Future of AI Is Small and On-Device

In this special holiday-weekend episode, Brian interviews Gary Ross, Associate Professor in the Faculty of Pharmaceutical Sciences at Kanazawa University, where he is Director of English and develops the online ESL program. Gary is a full-stack developer specializing in AI and online learning. He is the creator of J-Talk.com and the superb conference platform Edzilla, which we continue to use as our JALTCALL conference hosting system. We discuss Gary's various Kaken projects, including (1) Speech recognition in medical communication and training using ML with Jeanette Dennison, and (2) the Extensive Reading Artificial Intelligence app in collaboration with Mark Breirley https://gamma.erai.app/dashboard/manuscripts 
Check out Gary's J-Talk website for academics working in Japan(ese). https://j-talk.com/convert
and contact us at the podcast using jaltcall@jaltcall.org 
Contact Gary Ross at gary@p.kanazawa-u.ac.jp Kaken Researcher Number 10708142

Gary Ross Interview by Brian Gallagher Oct 2025

Heavybit is the leading early-stage investor in enterprise infrastructure. Since 2013, we've helped launch and scale visionary technical startups—from DevSecOps and feature flagging to AI code generation and beyond. We go hands-on from day 0, backing founders as they turn code into companies and build the future of software from the bottom up to the top down.

Unlock the secrets of data mastery with Shailendra, a 27-year veteran in AI, ML, and Data Science, driving AU$ 6 Billion in transformative value. As the author of 'Making Money out of Data,' their strategies redefine digital success. With expertise honed at IBM and Accenture, they drive innovation and data monetization. Explore how loyalty programs like Flybuys and Everyday Rewards revolutionize business—connect with Shaily today.

Japan association for Language Teaching (JALT) Computer Assisted Language Learning Special Interest Group. (JALTCALL)

Wei-Shen Wong and Anthony Malakian discuss the hottest financial technology topics in the capital markets every week.

New tracks tagged #Machine learning