Artwork

Content provided by Todd Ross Nienkerk. All podcast content including episodes, graphics, and podcast descriptions are uploaded and provided directly by Todd Ross Nienkerk or their podcast platform partner. If you believe someone is using your copyrighted work without your permission, you can follow the process outlined here https://ppacc.player.fm/legal.
Player FM - Podcast App
Go offline with the Player FM app!

Episode 53 – Language Biases in AI with Greg Bennett Pt. 2

19:16
 
Share
 

Manage episode 450184973 series 3615631
Content provided by Todd Ross Nienkerk. All podcast content including episodes, graphics, and podcast descriptions are uploaded and provided directly by Todd Ross Nienkerk or their podcast platform partner. If you believe someone is using your copyrighted work without your permission, you can follow the process outlined here https://ppacc.player.fm/legal.

Discover the hidden biases in large language models and how they skew towards English, creating unique challenges for other languages when it comes to artificial intelligence (AI). In this episode, Greg Bennett, head of prompt engineering at ZEALS, dissects the influence of training data predominantly sourced from English materials and the subsequent effectiveness of English prompts. We reveal how tiny linguistic tools called discourse markers can significantly enhance conversational flow in AI-generated responses, especially when navigating non-English languages. Linguistics professionals play a critical role in prompt engineering; their expertise in language structures and sociolinguistics gives them an edge in creating natural and engaging AI outputs. Hear about the intriguing differences between OpenAI and Anthropic's language models and how each handles discourse markers. We discuss why Anthropic's models seem to have an inherent grasp of these markers, whereas OpenAI's models need explicit instructions.

Reflecting on the evolution of conversational AI, we go back to the early days of voice assistants like Cortana in 2014, sharing personal experiences and challenges faced in those formative years. Fast forward to the present and we illustrate how advancements in generative AI have transformed conversational UX. Learn how specifying discourse formats, whether for articles or chat messages, can guide AI to produce the desired style and register, making interactions smoother and more intuitive.

Connect with us!

About Four Kitchens

Your website needs to fulfill your mission and stand the test of time. We have a long history of helping teams like yours — those who educate, advocate, and inform — realize their website’s potential and make a positive change in the world.

At Four Kitchens, we make the web a better place to teach, learn, and advocate for all. Driven by our content expertise, technical excellence, and a shared purpose of doing good, we craft websites that help organizations fulfill their missions.

Click here to learn more about how Four Kitchens makes websites with love. For good.

  continue reading

61 episodes

Artwork
iconShare
 
Manage episode 450184973 series 3615631
Content provided by Todd Ross Nienkerk. All podcast content including episodes, graphics, and podcast descriptions are uploaded and provided directly by Todd Ross Nienkerk or their podcast platform partner. If you believe someone is using your copyrighted work without your permission, you can follow the process outlined here https://ppacc.player.fm/legal.

Discover the hidden biases in large language models and how they skew towards English, creating unique challenges for other languages when it comes to artificial intelligence (AI). In this episode, Greg Bennett, head of prompt engineering at ZEALS, dissects the influence of training data predominantly sourced from English materials and the subsequent effectiveness of English prompts. We reveal how tiny linguistic tools called discourse markers can significantly enhance conversational flow in AI-generated responses, especially when navigating non-English languages. Linguistics professionals play a critical role in prompt engineering; their expertise in language structures and sociolinguistics gives them an edge in creating natural and engaging AI outputs. Hear about the intriguing differences between OpenAI and Anthropic's language models and how each handles discourse markers. We discuss why Anthropic's models seem to have an inherent grasp of these markers, whereas OpenAI's models need explicit instructions.

Reflecting on the evolution of conversational AI, we go back to the early days of voice assistants like Cortana in 2014, sharing personal experiences and challenges faced in those formative years. Fast forward to the present and we illustrate how advancements in generative AI have transformed conversational UX. Learn how specifying discourse formats, whether for articles or chat messages, can guide AI to produce the desired style and register, making interactions smoother and more intuitive.

Connect with us!

About Four Kitchens

Your website needs to fulfill your mission and stand the test of time. We have a long history of helping teams like yours — those who educate, advocate, and inform — realize their website’s potential and make a positive change in the world.

At Four Kitchens, we make the web a better place to teach, learn, and advocate for all. Driven by our content expertise, technical excellence, and a shared purpose of doing good, we craft websites that help organizations fulfill their missions.

Click here to learn more about how Four Kitchens makes websites with love. For good.

  continue reading

61 episodes

All episodes

×
 
Loading …

Welcome to Player FM!

Player FM is scanning the web for high-quality podcasts for you to enjoy right now. It's the best podcast app and works on Android, iPhone, and the web. Signup to sync subscriptions across devices.

 

Quick Reference Guide

Copyright 2025 | Privacy Policy | Terms of Service | | Copyright
Listen to this show while you explore
Play