Listen to the audio version of the article
Dear chatbot, help: I don’t know what to buy my sister for Christmas; she is almost forty years old, she loves cats, medicine, the gym. And artificial intelligence came to the rescue. More or less. This is the first Christmas where ordinary people are using artificial intelligence to help them with gifts. A first survey comes from Adobe, at the end of November, with as many as 30 percent of US users declaring they do so.
We tried Microsoft Copilot (Gpt4) and Bard (equipped with the new Gemini framework, launched in December) to help us with gifts for sister, mother and – since we were there – myself.
We do not recommend using the free ChatGpt because, as it is not connected to the internet, it cannot give updated suggestions on new models and current prices.
But some ideas can actually come from chatbots. Not only that: they can help us understand which specific products to choose, citing prices and features, and telling us which factors to consider depending on the profile of the person we want to give something to.
Are we at the beginning of a new way of giving gifts? Probably, if you consider that the first chatbots dedicated to shopping are arriving: Microsoft Shopping, Etsy, Shopify, which have begun to integrate them into their platforms (not yet for the Italian public). Google offers similar services (experimentally, on a group of selected accounts); ditto Amazon, which is also preparing to integrate a chatbot to guide users in purchasing. OpenAi has even launched SantaGpt, a customized version of ChatGpt Plus (for those who are subscribers) where you simulate a conversation with Santa Claus.
But how do current Chatbots fare? Those accessible to the Italian public, i.e. not yet well integrated with shopping tools, are useful even if not complete.
We started with the sister.
Bard began with generic advice, which seems aimed at a generic woman. Even some might find them sexist: “you could give her a piece of jewelry with her name on it, a painting or photograph of the two of you together, or a cookbook with her favorite recipes.” “You could give her a new kitchen appliance, a home accessory, or a beauty or personal care product.”
We pressed him, asking to be more specific, and he gave us gift ideas for cats: “A new toy or scratching post for your cat. A blanket or pillow with a cat pattern. A book or film about cats.” We asked for specific products and he gave us names, with a price range.
Ditto for mom. She also loves cats, as well as TV series and consumer electronics. There are also tips like “Consider the size and age of your mother’s cats. A scratching post for an adult cat may be too large or dangerous for a kitten.” As far as we’re concerned: “which running headphones should I buy? I listen to a lot of podcasts, but not music, when I run.”
The advice actually convinced us little. They feel like generic headphones and that Bard didn’t really follow our custom directions.
Copilot fared better. He indicated three specialized headphones for running, with prices, three lines of description for each, also explaining why they are good for that purpose.
We also asked for a summary of the reviews on a specific product and they cited three well-known American specialized magazines; has also added a series of sponsored links for direct purchase. A case of useful advertising.
Other things we can ask for: a summary of user reviews; what are the factors to consider when purchasing a certain product for a certain person; the pros and cons; If you can recommend alternatives, perhaps cheaper ones.
Copilot also gave us more targeted advice for our mother. For example, a cat-shaped heating pad (with the price: 19.99 euros) which doesn’t seem like a bad idea.
There may be errors, so the advice – as with all chatbot results – is to check with the original source. The chatbot can give wrong prices or get confused between the functions of different versions of the same product (which perhaps have similar names).
The future, in the short term, is chatbots integrated into shopping apps or platforms. In the USA it is already a reality.
Shopify – which claims to be a partner of millions of retailers around the world – was the first: in early 2023 it integrated OpenAI technology for its shopping assistant. It introduced an updated version on its website in October. We can ask him to help us find toys for a niece who is passionate about music and he shows other instruments suitable for children, immediately available for purchase with just a few clicks. Similar, but with more features, Copilot by Microsoft Shopping, which arrived in November and unfortunately is not yet available from us. It also does comparisons between products.
The tsy marketplace is testing a similar AI-powered search bot.
There’s also Google. For a few days it has had a specific function in Search Labs resources, however accessible only to a few accounts. When questioned, it shows suitable products, with useful links.
In December, a tool focused on clothing will also arrive in Google’s Search Labs which will allow you to design a dress using an AI image generator and will then direct you to similar products that actually exist.
Amazon hasn’t added generative AI to its search engine but has started using it to summarize product reviews for shoppers in the United States. AI-generated text that appears on product detail pages in the app and website highlights reviews from verified purchases.
In short: the usefulness of these tools is to give personalized ideas and cues; but also to advise us on the best based on the circumstances, summarizing the resources available online (reviews, prices, product sheets…). The usefulness will increase as chatbots become personalized, all-round shopping assistants, increasingly integrating multimodal functions (images, videos) and direct links to purchase.