Sign up to save your podcasts
Or
Share AI at Rightmove
Share to email
Share to Facebook
Share to X
Share to email
Share to Facebook
Share to X
Or
AI at Rightmove
The Rightmove Tech Podcast brings you news, views, and interviews from the Rightmove tech team.
Episode 5: AI at Rightmove
Welcome back to the Rightmove Tech podcast! In this episode, Ana and Chris speak with Elliot, Director of Technology, and Ayca, Senior Data Scientist, about AI. They talk us through the exciting ways in we are exploring AI solutions at Rightmove, both internally and externally, as well as career pathways and deeper insights into how these models work.
Listen in to find out more!
Chapter I – RAID at Rightmove
Elliot talks about the RAID project, which stands for Rightmove AI Discovery. This is a project to explore potential use cases for AI across Rightmove, be it for property hunters, Rightmove partners or internal engineering teams.
I do think the right strategy at this point is to invest in AI, and it’s about taking advantage of its current capabilities, but also creating a sort of readiness to take advantage of what’s around the corner.
Chapter II – How is AI changing Rightmove’s offering to partners and consumers?
There are several exciting potential use cases for AI, including (but not limited to) improving the core search experience for property-hunters and offering AI-boosted keyword search possibilities.
At this point, I see Rightmove offering a boost to existing functionality rather than creating complete new experiences. But over time, I think that’s going to change, particularly with images or video.
Chapter III – How is AI changing how the Rightmove tech team works?
Elliot talks about how we are already using GitHub Copilot in the engineering division and plans to support engineers with learning the new AI skills they will need.
I think it’s going to be something that everyone really needs to have some good skills in. So we need to make sure that we help people with that. It’s important to help people find the right work, that’s going to keep them interested and help them develop.
Chapter IV – Ayca’s Journey into AI
Chris and Ayca talk about how she got into the AI field; she started with a Bachelor’s degree in Statistics and then went on to do a Master’s in Applied Statistics. Her first role was as a Data Analyst, and over her career she has led several AI and machine learning projects at different companies.
Ayca highlighted that strong programming skills are a must for those looking to break into this field, with Python being the go-to language for many in AI.
This role allows me to engage with innovative projects and apply strategic thinking, all within a supportive environment where leadership truly appreciates the significance of data and innovation.
A career in AI is not just about handling data and building models, but about understanding how this data works towards making meaningful outcomes. It’s about seeing the bigger picture.
Chapter V – Large Language Models and how they evolved
In this section, Ayca sheds fascinating light on the subject of Large Language Models (LLMs) and how they evolved from older models like Word2Vec and FastText. While these two methods relied on static word embeddings (a high dimensional vector representation of a word), modern Transformers can dynamically generate embeddings based on the context of a word in a sentence – in her own words, a game changer! Transformers can be trained on much larger data sets because they can process all words in a sentence simultaneously and can learn patterns about how language is used in different contexts.
At the heart of Transformers is the attention mechanism, which allows the model to focus on different parts of the input sentence when predicting each word.
Ayca acknowledges that choosing the right model is a challenging problem, given the vast amount of AI models currently out there. You should assess the model against the right benchmarks, such as RUGE or GLUE, depending on what you want to achieve.
Chapter VI – Understanding how these models work (or sometimes don’t)
Models […] process vast amounts of unstructured text. They learn by predicting the next word in sentences over and over again on a massive scale. By doing this, they learn statistical relationships between sequences of words and the context in which they appear.
Ayca and Chris also delved into some examples of the problems that can occur when a model hasn’t been trained successfully and suggestions for how to overcome these challenges.
Aaaaand that’s a wrap for Season 1 of the Rightmove Tech podcast! Thank you for joining us on this journey – stay tuned for news on Season 2 and, as always, thank you for listening.
Credits
Produced by Ana and Chris.
Edited by Chris.
Reviewed and published by Appy.
Artwork by Tay.
The post Podcast: AI at Rightmove appeared first on Rightmove Tech Blog.
View all episodes
By Rightmove Tech PodcastThe Rightmove Tech Podcast brings you news, views, and interviews from the Rightmove tech team.
Episode 5: AI at Rightmove
Welcome back to the Rightmove Tech podcast! In this episode, Ana and Chris speak with Elliot, Director of Technology, and Ayca, Senior Data Scientist, about AI. They talk us through the exciting ways in we are exploring AI solutions at Rightmove, both internally and externally, as well as career pathways and deeper insights into how these models work.
Listen in to find out more!
Chapter I – RAID at Rightmove
Elliot talks about the RAID project, which stands for Rightmove AI Discovery. This is a project to explore potential use cases for AI across Rightmove, be it for property hunters, Rightmove partners or internal engineering teams.
I do think the right strategy at this point is to invest in AI, and it’s about taking advantage of its current capabilities, but also creating a sort of readiness to take advantage of what’s around the corner.
Chapter II – How is AI changing Rightmove’s offering to partners and consumers?
There are several exciting potential use cases for AI, including (but not limited to) improving the core search experience for property-hunters and offering AI-boosted keyword search possibilities.
At this point, I see Rightmove offering a boost to existing functionality rather than creating complete new experiences. But over time, I think that’s going to change, particularly with images or video.
Chapter III – How is AI changing how the Rightmove tech team works?
Elliot talks about how we are already using GitHub Copilot in the engineering division and plans to support engineers with learning the new AI skills they will need.
I think it’s going to be something that everyone really needs to have some good skills in. So we need to make sure that we help people with that. It’s important to help people find the right work, that’s going to keep them interested and help them develop.
Chapter IV – Ayca’s Journey into AI
Chris and Ayca talk about how she got into the AI field; she started with a Bachelor’s degree in Statistics and then went on to do a Master’s in Applied Statistics. Her first role was as a Data Analyst, and over her career she has led several AI and machine learning projects at different companies.
Ayca highlighted that strong programming skills are a must for those looking to break into this field, with Python being the go-to language for many in AI.
This role allows me to engage with innovative projects and apply strategic thinking, all within a supportive environment where leadership truly appreciates the significance of data and innovation.
A career in AI is not just about handling data and building models, but about understanding how this data works towards making meaningful outcomes. It’s about seeing the bigger picture.
Chapter V – Large Language Models and how they evolved
In this section, Ayca sheds fascinating light on the subject of Large Language Models (LLMs) and how they evolved from older models like Word2Vec and FastText. While these two methods relied on static word embeddings (a high dimensional vector representation of a word), modern Transformers can dynamically generate embeddings based on the context of a word in a sentence – in her own words, a game changer! Transformers can be trained on much larger data sets because they can process all words in a sentence simultaneously and can learn patterns about how language is used in different contexts.
At the heart of Transformers is the attention mechanism, which allows the model to focus on different parts of the input sentence when predicting each word.
Ayca acknowledges that choosing the right model is a challenging problem, given the vast amount of AI models currently out there. You should assess the model against the right benchmarks, such as RUGE or GLUE, depending on what you want to achieve.
Chapter VI – Understanding how these models work (or sometimes don’t)
Models […] process vast amounts of unstructured text. They learn by predicting the next word in sentences over and over again on a massive scale. By doing this, they learn statistical relationships between sequences of words and the context in which they appear.
Ayca and Chris also delved into some examples of the problems that can occur when a model hasn’t been trained successfully and suggestions for how to overcome these challenges.
Aaaaand that’s a wrap for Season 1 of the Rightmove Tech podcast! Thank you for joining us on this journey – stay tuned for news on Season 2 and, as always, thank you for listening.
Credits
Produced by Ana and Chris.
Edited by Chris.
Reviewed and published by Appy.
Artwork by Tay.
The post Podcast: AI at Rightmove appeared first on Rightmove Tech Blog.