AI Agents / Agentic AI: The next generation of AI?

03/10/25 • 19 min

Embedded AI - Intelligence at the Deep Edge

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 a text

Support the show

Previous Episode

LLMs - Fancy Autocorrect or can they actually Reason?

Send us a text

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

Next Episode

AI Bias and Fairness: A Contentious Landscape

Send us a text

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