The Clearly Podcast
Machine Learning Pt1 - What is Machine Learning Anyway?
Summary
As a big finale to Season 2, we thought we'd bring you a mini sub-series on possibly the biggest buzz word of the year: Machine Learning.
This week, we look at what machine learning is, and where you might have come across it in your day to day life. We go on to discuss why business might want to use Machine Learning and the sorts of problems it can solve.
Next week we will go on to talk about the considerations you will need if you want to use Machine Learning and in the final episode we look at using ML components inside Power BI and other tools.
Artificial Intelligence in a more general sense is also the topic of this year's Reith Lectures. Just to prove how important and topical The Clearly Podcast really is.
If you already use Power BI, or are considering it, we strongly recommend you join your local Power BI user group here.
Transcript
Andy: Hello everyone! How are we today?
Shailan: Feeling carbon offset neutral.
Tom: Very well, thanks.
Andy: Let's dive into machine learning. This is the first of three discussions. Today, we'll define machine learning, a term that's often used without context. So, Tom, what is machine learning?
Tom: Machine learning is a set of tools and algorithms applied to datasets to make inferences. These inferences can then improve the model. Essentially, it's about making predictions from data, moving from just reporting to predictive and prescriptive analytics.
Andy: Where is machine learning used today?
Shailan: We encounter machine learning frequently, like in retail with basket analysis suggesting related products. Companies like Amazon and Netflix use it extensively. Social media also uses it to tailor ads and content to users.
Andy: There is backlash, especially in social media, where the content can affect mental health. Why is machine learning such a buzzword now?
Tom: The availability of tools and increased compute power has made machine learning more accessible. Cloud computing allows for large datasets to be processed without significant infrastructure costs. However, there's a misconception that AI can solve all problems, which isn't true.
Shailan: Automated customer service can be frustrating, highlighting that AI isn't always the best solution.
Tom: Exactly. Cost reduction through AI can lead to poor customer service. AI needs to be used thoughtfully.
Shailan: All platforms have AI capabilities, but the effectiveness depends on the data quality and the specific application.
Tom: Bias in datasets can lead to biased AI. For example, if training data is biased, the AI will be too. This is a critical issue.
Andy: Recent phone releases have improved how cameras handle different skin tones, showing progress in addressing bias. Machine learning isn't a one-size-fits-all solution.
Tom: Machine learning needs to be applied thoughtfully, especially in niche areas like medical science and public health, where it can make a significant positive impact.
Andy: Machine learning models in fields like oil exploration have shown great promise. But remember, ML isn't the answer to every problem. It's used effectively in retail and media, but its true potential lies in niche, targeted applications.
Shailan: Defense is another area where machine learning is used, such as in explosive device detection and training simulations.
Tom: Machine learning complements traditional tools like reporting and dashboarding but doesn't replace them. It's important to use the right tool for the right job.
Andy: This is the first of three podcasts on machine learning. Next, we'll discuss when organizations might be ready to adopt machine learning, and finally, we'll look at how machine learning integrates with tools like Power BI and Excel.
Tom: Looking forward to it. And yes, we will talk about how machine learning can enhance Excel.
Shailan: Absolutely. Stay tuned!