
Sign up to save your podcasts
Or


This episode of Techsplainers explores data governance, the essential framework that ensures organizational data is properly managed, protected, and utilized. Amanda explains how data governance serves as an "air traffic control system" for information, defining policies and procedures for data collection, storage, and usage throughout its lifecycle. The discussion covers the four key components of governance frameworks: program goals and roles, data standards and policies, auditing procedures, and supporting tools. We examine how effective governance delivers tangible benefits including enhanced data value, balanced access, compliance with regulations like GDPR and HIPAA, and responsible AI development. The episode also addresses common implementation challenges such as lack of sponsorship, inconsistent architecture, and evolving AI requirements, before concluding with best practices including automation, creating a comprehensive data catalog, and continuous improvement. As the final installment in our data for AI series, this episode demonstrates how governance provides the structure that enables everything from AI-ready data to synthetic data creation.
Find more information at https://www.ibm.com/think/topics/unstructured-data
Find more episodes at https://www.ibm.biz/techsplainers-podcast
Narrated by Amanda Downie
This episode of Techsplainers explores synthetic data - artificially generated information designed to mimic real-world data while preserving statistical properties and patterns. Amanda explains how synthetic data has become critical for AI development by addressing issues of data scarcity, privacy concerns, and training needs. The discussion covers the three types of synthetic data (fully synthetic, partially synthetic, and hybrid) and various generation techniques including statistical methods, GANs, transformer models, VAEs, and agent-based modeling. We examine the significant benefits of synthetic data - customization flexibility, improved efficiency, enhanced privacy protection, and data enrichment - while also addressing challenges like bias propagation, model collapse, accuracy-privacy tradeoffs, and verification needs. The episode concludes with real-world applications across automotive, finance, healthcare, and manufacturing industries, demonstrating how synthetic data is becoming essential for AI development.
Find more information at https://www.ibm.com/think/topics/unstructured-data
Find more episodes at https://www.ibm.biz/techsplainers-podcast
Narrated by Amanda Downie
This episode of Techsplainers explores unstructured data - information without predefined formats that makes up 90% of enterprise data. Amanda explains how unstructured data differs from structured and semi-structured data, covering its diverse sources from emails to social media posts to sensor data. The discussion highlights why unstructured data has transformed from "dark data" into a strategic asset, particularly for AI applications. We explore key use cases including generative AI training, retrieval augmented generation (RAG), sentiment analysis, and predictive analytics. The episode also covers storage solutions like object storage and data lakes, plus processing tools that help organizations extract value from their unstructured information. With proper governance and management, unstructured data has become the fuel powering today's AI revolution.
Find more information at https://www.ibm.com/think/topics/unstructured-data
Find more episodes at https://www.ibm.biz/techsplainers-podcast
Narrated by Amanda Downie
This episode of Techsplainers explores the concept of "bad data" - information that compromises decision-making because it's inaccurate, incomplete, inconsistent, outdated, duplicate, invalid, or biased. We examine why bad data is particularly dangerous due to its stealthy nature, often going undetected until significant damage occurs. Through real-world examples like Unity Technologies' $110 million loss from bad data in their AI algorithms, we illustrate the severe consequences across industries from healthcare to finance. The discussion covers the diverse causes of data quality problems - from system failures and data decay to human error and integration challenges - and provides a comprehensive approach to prevention through governance, monitoring, cleansing, and data literacy. As organizations increasingly rely on AI systems, understanding that "garbage in, garbage out" applies more than ever becomes crucial for success in data-driven initiatives.
Find more information at https://www.ibm.com/think/topics/bad-data
Find more episodes at https://www.ibm.biz/techsplainers-podcast
Narrated by Amanda Downie
From the publisher's feed