our views and our knowledge in analytics and other releveant topics

our blogs

Breaking down AI


Artificial Intelligence is definitely the current buzz word in all things digital and, after spending two days at the recent AI Summit, I’ve realised that AI can be explainable; It isn’t just about computers beating humans at chess games, nor industries moving towards a robotic workforce.

So I’m a convert. AI isn’t as big and scary as first thought and it’s not going to steal my job…yet. It’s actually at a really interesting developmental place where it can help digital analysts lighten their workload, be more efficient and perform deeper analyses. Win-win really.

Interestingly, a lot of us are probably already using some type of AI or machine learning, we just might not be aware of it. For example, if you use AB testing or recommendation engines, price optimization software or Google – these types of tools have developed parts of their offer to include AI. There’s a machine learning algorithm that is working to decide what people should see or be exposed to.

There’s more to the advancement of AI than relying on tools or systems that are already using it. Here is my STOP, START, CONTINUE advice for getting you AI ready (and using):



…thinking of AI as something completely different to what you’re already doing. Instead think of it as an enhancement. AI capabilities are already becoming normalized in some areas of work and everyday life. Positioning it as an alien concept or something that is going to radically revolutionise analytics in an instant is wrong. Instead, it’s likely to be a “quiet revolution”, one that slowly increments advancements rather than a paradigm shift.



…getting involved in writing your own AI algorithms. Both R and Python have extensive documentation online that will get you up to speed with tried and tested AI algorithms. TensorFlow is a great place to start and has some good open-source projects available.  Alternatively, there are specific data science platforms that can help you develop, prototype and run AI code.



…interrogating the current tools you use. Think with an AI frame of mind, if something could work better with an AI algorithm challenge your providers to develop it.