EP91: Dr Ahmed El Adl, Executive Advisor: Vulnerable AI and the unintended weaknesses in algorithms

EP91: Dr Ahmed El Adl, Executive Advisor: Vulnerable AI and the unintended weaknesses in algorithms
10 de ago. de 2020 · 47m 49s

💬 “10 years ago everyone talked about big data, how to collect it, store it, secure it, manage it and use it. Today we are not talking about big data...

mostra más
💬 “10 years ago everyone talked about big data, how to collect it, store it, secure it, manage it and use it. Today we are not talking about big data analytics. We are talking about how to gain the most insight from the data, which is a different world. It will see businesses move from analysing big data with machine learning to working out how they can apply reasoning capabilities to drive insights from smaller, but more useful data.”

This is a conversation with Dr Ahmed El Adl.

🎙️ Ahmed is a technology and business innovation leader with a proven track record of global scale achievements in envisioning and building intelligent technologies and applying them to solve real-world problems to accelerate the digital and cognitive transformation of businesses, workforce, services, and machines. His main focus is the right industrial and business adoption of AI/ML, IIoT, and the concept of Cognitive Digital Twin (CDT). Ahmed has Ph.D. and Post Doc. in Computer Science with a focus on Artificial Intelligence and Industrial Robotics.

🎧 In this episode, Ahmed explains what it takes to make machine learning algorithms safe and secure, but also how very simple changes to external stimuli, for example in confusing computer vision can lead to at best silly errors and at worst disasters.

For example, Ahmed explains how a rogue actor that puts a heat source near to a sensor in a power plant, could trigger the algorithm into shutting down the plant. He goes on to explain how it is possible to mitigate for this.

Ahmed goes on to explain that currently we are still having to hard code reasoning logic into machines and AI projects fail because project architects can’t connect the different systems and layers, not because the machine learning algorithms weren’t good enough.

Ahmed talks about how valuable TikTok’s data will be to Microsoft and how they may use it if they acquire the platform’s North American user base.

Finally, we come full circle on the point of the episode, that business opportunities with be uncovered when businesses move away from the big data and analytics mindset to the insight mindset. From analytic and big data and machine learning to reasoning.
mostra menos
Información
Autor Richard Foster-Fletcher
Página web -
Etiquetas

Parece que no tienes ningún episodio activo

Echa un ojo al catálogo de Spreaker para descubrir nuevos contenidos.

Actual

Portada del podcast

Parece que no tienes ningún episodio en cola

Echa un ojo al catálogo de Spreaker para descubrir nuevos contenidos.

Siguiente

Portada del episodio Portada del episodio

Cuánto silencio hay aquí...

¡Es hora de descubrir nuevos episodios!

Descubre
Tu librería
Busca