Artificial Intelligence, once the reserve of highly advanced robotics or backroom geeks, has now entered the business vocabulary across every sector and organisations, including small businesses like us, are starting to investigate the benefits (and downsides) of using AI.
While it’s a sexy topic with limitless potential, it can throw up some interesting questions for market researchers. So, rather than jump in with both feet, we decided to explore AI with one component of our research work: coding of open-ended questions.
Although advances in AI - natural language processing and machine learning - have made it possible to automate the analysis of open-ended questions, up until recently it wasn’t economically viable to develop and operate cutting-edge deep learning algorithms for an area as niche and complex, with relatively small sample sizes, as healthcare market research.
However, all of that changed a few months back when we discovered an overseas software program specifically designed for market researchers. We were attracted to this program because the benefits were two-fold:
1. For our clients: deeper insights, more relevant and actionable results
2. For our team of researchers: AI-assisted text analysis augmenting current practices, leading to more efficiencies in coding
So far, we are still in trial phase with this program. However, we have already made some interesting distinctions and gained a clearer understanding about when and why AI is our ‘friend’, and when it’s our ‘foe’.
AI as our ‘friend’.
- Time taken for analysis of certain types of open-ended questions has been reduced significantly – what was taking hours can now be done in minutes
- ‘AI meets human’ seems to be the best way of making this useful and valuable – AI is excellent for reliably executing codes that it has ‘learnt’ or ‘been trained on’, whilst a human mind’s flexibility allows us to understand context and adjust codes as required
AI as our ‘foe’.
- When we started out, we had probably underestimated how much human input would still be necessary when coding. We are discovering that the more we use the program, the less input is needed from us as the machine ‘learns’, but we are still definitely in the teaching phase of working with this AI
- We have also realised (which is also a relief!) that AI will never fully replace what a trained, insightful human can do. AI leads us to faster insights and state-of-the-art text analysis, but there is nothing like the human eye and human mind to see ‘beauty’ in data that a machine will never appreciate.
Overall, we are very optimistic about what we can achieve with AI in the future. It’s a major new presence in our industry and we believe it will enhance what we can provide to clients as well as making our own lives easier.
If you are interested in how we can apply AI to your next study, please email us for a no-obligation chat on info@sowhatresearch.com.au