Embedded AI - Intelligence at the Deep Edge

Send us Fan Mail

In this episode, we dive into the fascinating convergence of blockchain and artificial intelligence (AI)—two powerful technologies reshaping the digital landscape. We explore how blockchain brings transparency, trust, and data integrity to AI systems, while AI enhances blockchain networks through automation, prediction, and optimization.

You’ll hear about real-world applications across industries like finance, healthcare, and smart infrastructure, along with an overview of pioneering platforms such as SingularityNET and Ocean Protocol, which are creating decentralized marketplaces for data and AI models. We also discuss the promise of auditable AI, where blockchain provides an immutable record of AI decision-making.

Of course, the episode doesn’t shy away from the challenges, including scalability issues, complexity, and ethical concerns. But the overarching theme is clear: the fusion of blockchain and AI has the potential to create more secure, intelligent, and decentralized systems—paving the way for the next generation of digital innovation.

Support the show

If you are interested in learning more then please subscribe to the podcast or head over to https://medium.com/@reefwing, where there is lots more content on AI, IoT, robotics, drones, and development. To support us in bringing you this material, you can buy me a coffee or just provide feedback. We love feedback!

Send us Fan Mail

In this episode, we examine the critical issue of bias in artificial intelligence, exploring how biased AI systems can amplify discrimination and perpetuate societal inequalities. We discuss the sources of AI bias, including prejudiced training data, algorithmic design choices, and human decisions during development. We highlight how biased AI impacts areas like recruitment, criminal justice, healthcare, finance, and social media, potentially deepening existing inequalities and undermining public trust.

We also delve into efforts to address AI bias through technical solutions—such as collecting diverse data and using fairness-oriented algorithms—as well as regulatory responses like the EU AI Act and emerging legislation in the United States. Yet, despite these efforts, defining and effectively mitigating AI bias remains a significant challenge. Ultimately, we emphasize the importance of interdisciplinary collaboration and ethical guidelines to ensure AI systems are fair, equitable, and trustworthy.

Support the show

If you are interested in learning more then please subscribe to the podcast or head over to https://medium.com/@reefwing, where there is lots more content on AI, IoT, robotics, drones, and development. To support us in bringing you this material, you can buy me a coffee or just provide feedback. We love feedback!

Send us Fan Mail

In this podcast episode, we investigate AI agents—autonomous systems that sense their environments, make independent decisions, and carry out tasks. We discuss their various types, architectures, and capabilities, highlighting their limitations and ethical implications. Special attention is given to the rise of Agentic Workflows and Data Synthesis, driven by challenges around accuracy in current AI systems. The episode also explores practical advice on building effective agents, emphasizing iterative prompt engineering and standardized JSON outputs. Finally, we touch on Edge AI Agents, a promising area bringing autonomous intelligence directly to resource-constrained devices, shaping the future of AI applications.

Support the show

If you are interested in learning more then please subscribe to the podcast or head over to https://medium.com/@reefwing, where there is lots more content on AI, IoT, robotics, drones, and development. To support us in bringing you this material, you can buy me a coffee or just provide feedback. We love feedback!

Send us Fan Mail

In this episode, we discuss the limitations of Large Language Models (LLMs) in areas like deductive reasoning, analogy-making, and ethical judgment. While today’s AI models excel at recognizing statistical patterns in vast datasets, they lack genuine understanding or an internal model of the world. Researchers are tackling these challenges through innovations such as causal AI, inference-time computing, and neuro-symbolic approaches, all aimed at enabling AI to move beyond mere pattern recognition towards true reasoning.

We explore how these emerging technologies, including causal inference, inference-time computing, and neuro-symbolic integration, are pushing AI closer to human-like, “System 2” reasoning. Will these advancements finally bridge the gap between AI imitation and genuine reasoning? Tune in as we dive into the future of artificial intelligence and explore what it will take for machines to truly think.

Support the show

If you are interested in learning more then please subscribe to the podcast or head over to https://medium.com/@reefwing, where there is lots more content on AI, IoT, robotics, drones, and development. To support us in bringing you this material, you can buy me a coffee or just provide feedback. We love feedback!

Send us Fan Mail

In this episode, we explore the unique nature of AI mistakes and why they differ fundamentally from human errors. Unlike people, AI systems make random, inconsistent, and unpredictable errors, often without awareness of their own limitations. This unpredictability challenges traditional security approaches, requiring new frameworks for AI reliability and risk management.

The discussion delves into two potential solutions: engineering AI to make more human-like mistakes and creating specialized mistake-correcting mechanisms tailored for AI. While AI can exhibit human-like behaviors—such as prompt sensitivity and biases learned from training data—it also introduces distinct vulnerabilities that require fresh security strategies.

How can we ensure AI is deployed safely in decision-making? And what do these insights mean for the future of AI security? Tune in for an eye-opening conversation on the evolving landscape of AI safety and reliability.

Support the show

If you are interested in learning more then please subscribe to the podcast or head over to https://medium.com/@reefwing, where there is lots more content on AI, IoT, robotics, drones, and development. To support us in bringing you this material, you can buy me a coffee or just provide feedback. We love feedback!

Send us Fan Mail

In this episode, we explore Apple’s strategic partnership with Alibaba, integrating the Qwen AI model into iPhones sold in China. Faced with regulatory barriers and declining sales, Apple turns to Alibaba’s powerful large language model (LLM) to bring advanced AI features to its Chinese users while ensuring compliance with local laws.

We break down the implications of this move—how it strengthens Apple’s foothold in China, boosts Alibaba’s AI credibility, and reflects the broader trend of AI localization in global markets. However, challenges loom, including government scrutiny, competition from local AI firms, potential performance limitations, and privacy concerns.

Is this a smart strategic play or a risky compromise? And what does it mean for US-China tech relations? Tune in as we unpack the stakes behind Apple’s AI decision in China.

Support the show

If you are interested in learning more then please subscribe to the podcast or head over to https://medium.com/@reefwing, where there is lots more content on AI, IoT, robotics, drones, and development. To support us in bringing you this material, you can buy me a coffee or just provide feedback. We love feedback!

Send us Fan Mail

In this episode, we investigate the world of AI-generated music, exploring how cutting-edge AI techniques—such as transformers, GANs, and VAEs—are revolutionizing music creation. We also take a look at traditional non-AI methods like arpeggiators and Markov chains, which continue to shape algorithmic composition.

Beyond the software, we discuss an innovative MIDI controller design tailored for AI-driven music production, featuring controls specifically optimized for manipulating AI parameters and integrating seamlessly with modern music tools.

But with great innovation comes great debate—what are the ethical and legal implications of AI-generated music? We tackle concerns surrounding copyright, originality, and the potential impact on human musicians. Is AI enhancing creativity, or is it replacing it? Tune in for an insightful discussion on the future of music in the age of AI.

Support the show

If you are interested in learning more then please subscribe to the podcast or head over to https://medium.com/@reefwing, where there is lots more content on AI, IoT, robotics, drones, and development. To support us in bringing you this material, you can buy me a coffee or just provide feedback. We love feedback!

Send us Fan Mail

In this episode, we explore the concept of AI hard takeoff—the moment when artificial intelligence rapidly surpasses human intelligence, triggering an unstoppable acceleration in its capabilities. We break down the difference between a hard takeoff and a soft takeoff, weighing the potential risks and benefits of AI evolving beyond human control.

We also examine recent breakthroughs in AI, uncovering evidence that suggests we may be closer than ever to Artificial General Intelligence (AGI) and even Artificial Superintelligence (ASI). With insights from leading AI experts, we discuss the growing concerns over the speed of AI development and whether organizations and governments are truly prepared for what comes next.

Is humanity on the brink of an AI revolution, or are we rushing into unknown dangers? Tune in to find out.

Support the show

If you are interested in learning more then please subscribe to the podcast or head over to https://medium.com/@reefwing, where there is lots more content on AI, IoT, robotics, drones, and development. To support us in bringing you this material, you can buy me a coffee or just provide feedback. We love feedback!

Send us Fan Mail

In this episode, we investigate DeepSeek AI, the cost-effective yet high-performing Chinese AI model that is making waves in the industry. We compare its capabilities to leading American models like OpenAI’s, uncovering how DeepSeek achieves impressive reasoning and coding performance at a fraction of the training cost.

We break down the innovative techniques behind its success, including the Mixture-of-Experts architecture and Multi-Head Latent Attention mechanism, which contribute to its efficiency. But what does this mean for the global AI landscape? We explore the potential disruption to major tech companies, the implications of DeepSeek’s open-source nature, and the ripple effects on the US stock market and the future of AI development.

Is DeepSeek the beginning of a new AI era, or a major challenge to Western AI dominance? Tune in to find out!

Support the show

If you are interested in learning more then please subscribe to the podcast or head over to https://medium.com/@reefwing, where there is lots more content on AI, IoT, robotics, drones, and development. To support us in bringing you this material, you can buy me a coffee or just provide feedback. We love feedback!

Send us Fan Mail

This episode talks about the essentials of exploratory data analysis (EDA) for image recognition. We discuss key techniques—descriptive, diagnostic, and predictive EDA—and outline recommended steps such as image visualization, statistical analysis, anomaly removal, and feature engineering, along with ethical considerations in the process.

We also explore how EDA enhances model accuracy, focusing on the person detection model MCUNet-VWW2 and the Wake Vision dataset. Learn how label correction, data augmentation, and preprocessing improved performance while addressing dataset features, limitations, and the impact of EDA in real-world applications. Join us for an insightful guide to mastering EDA in image recognition!

Support the show

If you are interested in learning more then please subscribe to the podcast or head over to https://medium.com/@reefwing, where there is lots more content on AI, IoT, robotics, drones, and development. To support us in bringing you this material, you can buy me a coffee or just provide feedback. We love feedback!