This year, I have been repeatedly asked if it is a good idea to generate Christmas shopping lists using artificial intelligence (AI). Clearly it is convenient, but does it really come up with the best ideas for what to give near and dear ones?
The use of AI is becoming commonplace; take ChatGPT, for example, which currently has more than 300 million weekly “active” users. It is a large language model that is trained on extensive amounts of text data and effectively produces statistical patterns based on the data collected. It is important to remember that systems such as ChatGPT are based on previously harvested text data. Hence, it is not tailor-made to you as an individual, meaning it can never exactly predict what your favourite auntie may want as it produces a generic response. And often it generates many suggestions.
Choice overload
The fact that AI systems may produce a high number of possibilities is a potential problem as people are not usually good with too many choices. Generally, when faced with more than six product possibilities, people are more likely to walk away, and people also tend to be less satisfied with the choices they make. This is, at least partially, because consumers find it difficult to deal with all the information related to each product and, thus, results in their cognitive resources being overwhelmed. Hence, it is not surprising that when I asked ChatGPT what to buy for my grandma this Christmas and got 32 suggestions, I felt a little dazed. Maybe also because they all seemed rather random.
Source: Tara WInstead / Pexels
The user shapes the response
It is worth noting that the success of using AI systems is very much dependent on the user. The more information you provide, the better the answer it can give. Possibly, it can even help to ensure that people have a higher level of acceptance for a high number of options as it will slimline the answers and, thus, require less cognition to process. This has been found for the recommendations of songs. However, it is worth noting that songs are not necessarily the same as picking one item that you are going to give to a friend in that, with songs, having more variety and options means you can choose something that is more fitting for your mood, circumstances, etc.
Price and emotion as predictors of purchase
In many cases, price predicts what people buy. That is something else that AI systems struggle to consider, and, thus, the final decision may deviate from what suggestions that have been made. Another aspect that AI systems can’t consider is the emotions people feel for others. Such emotions can have a strong influence on what consumers purchase. Exploring different types of products may be driven by the emotions you have for others. But if you simply look at a list, you are likely to think rationally about what you may buy and, thus, exclude the emotional side. This may result in a purchase that may not reflect how you feel about another person.
The moral of the story?
While it may be tempting to rely on ChatGPT to help you with your Christmas present choices this year, it is probably better to take your time and think about what people you care about may really like.