First had a thought to write about ecommerce but then realized that the change is going to be not just about ecommerce but how we experience commerce in general. And the change I am talking about here is how AI changes so many things we are used to in the past.
Think of rather normal ecommerce activity, it starts with the consumer’s need to purchase. We are talking about need for something particular need, it could also urge to purchase random things but let’s focus on the intention to buy something first.
Consumer recognizes the need to purchase an item, e.g. bicycle in this example. Often the fulfillment starts with exploration of the options, reviews, recommendations, prices, sellers etc. Rather time-consuming, sometimes nerve-breaking (e.g. reading about contradictory reviews about the same item) and yes, could be also insightful to learn how other people are learning the best products for their needs.
Now, imagine if this could be done as an ordinary prompt for ChatGPT or Bing copilot:
“I am buying a mountain bike for semi-serious use, I am tall (190 cm), my budget is 1500 eur and I would be happy to receive the bike quickly (within few days), what are my options?”
What the consumer would expext already know if they are anyhow familiar with ChatGPT and alike that the AI bot would be able to start drafting up a shortlist about possible purchases, prices, availability and direct links to sites where to buy. And even make references to reviews.
We are not there yet but as everyone has seen the base of AI development lately it most likely will not take forever to land something similar that was just described above. ChatGPT is not connected (expect pro seems to be somehow) live datasets but testing above scenario with Bing copilot (closed beta at the moment), there was already pointers made available to possible purchases and elaborative questions for figure out better the needs the consumer has.
One significat bottleneck on this evolution is very likely the product data quality in general. After working years with extensive product datasets it is obvious that each industry sector is screaming for common datamodels or ways to a) structurize their own data and b) make it available for common use, be it then retail chains, search engines or AI/ML learning new datasets. Investing in data space design is probably one of the core things in IT industry for coming years if these challenges are somehow going to be addressed.
In the beginning of this post there was mention of commerce, not just digital side of it. Now imagine above scenario of bicycle purchase occuring offline, on retail store. That poor sales guy either competes with all the AI knowledge (as the consumer can access it on their mobile while being present by the product) or uses the same tactic; AI assisted sales where the humans are output/input for AI systems then finding the best fit.