ChatGPT and generative AI: A look under the hood

Our Data Scientist, Gabe Musker, has shared his views on ChatGPT and generative AI.

Gabe’s aim was to ‘look under the hood’ of generative AI tools such as ChatGPT, and provide our team, in non-technical language, an understanding of how machines learn, from the most basic algorithms all the way up to the cutting edge.

Our goal was to reduce the fear around conversations about AI, and give our team the tools they need to make informed decisions about safe and practical uses of modern AI.

So, what were some of our key takeaways? Read on to find out!

Pick the right AI for the job

AI tools can be extremely complex and clever, but at their core, they are just mathematical equations, designed to optimise one thing. The AI behind ChatGPT would be terrible at playing chess, or driving a Tesla, whereas others would excel at these. At its core, something like ChatGPT is just basically a really clever parrot – it can give convincing-sounding answers to text prompts, but that’s all it’s good for.

Does machine learning equal AI?

Most tools that use machine learning – a trial-and-error approach to optimisation, focused on learning a pattern via repeated exposure to a particular data set – can be called AI. However, the reverse is not necessarily true – tools like Deep Blue, the chess-playing algorithm that beat Gary Kasparov, did not use machine learning techniques, but is indisputably AI.

Don’t miss out on the easy wins

The over-focus on extremely complex, computation-heavy (and hence expensive) machine learning tools is causing businesses to miss out on the easy wins that simple AI tools could provide. At Branding Science, our focus is on creating minimal viable models, built and optimised rapidly, that can bring the most insight to our clients at the lowest cost and in the shortest amount of time.

Don’t get left behind

Understanding and harnessing data science methods, including AI & ML techniques, is no longer optional for us, our clients, or the market research industry. It will not be long before these techniques are the preferred way of understanding a market, customer segment, or business opportunity, and anyone who isn’t up to speed will be left behind, making worse decisions based on less insight than their competitors.