PG: That makes sense. As for the empathic AI, that was amazing, the idea of how much that technology has grown, that it really is approaching human-like levels. Have you any intelligence on how the grocery channel is using AI?
Rachel Wilkinson: There’s a very interesting company, and they have implemented an AI assistant that looks initially like it’s a digital screen, like a virtual assistant. How they’re using it is a bit like your favorite shop assistant, your trusted store assistant. It’s used by the store manager to inform customers on promotions, where they should go. It will help them deliver recipe ideas. So if someone says, “I’ve got 10 pounds, I need to buy a meal for my family tonight, what do you recommend?,” It will literally take you to the aisle. It looks like a person, a real person. And when they set this up, they scan the entire store so that it has a knowledge of the aisle, every single product within that store. And then when someone asks for something, it knows exactly what product and it has all the information about that product.
[RELATED: EXCLUSIVE: At CES 2024, What Retailers Need to Know About AI"]
The store managers are using it to keep an eye on what’s happening in the store and to monitor the questions, so it knows what its customers are interested in and not interested in; it’s collecting data. It’s quite useful for the buyers as well to negotiate with their vendors. So it’s very much like a front-end and a back-end AI system. It’s almost like the eyes and ears of what’s happening in the store. And, of course, in a grocer where lots of people come and go, and staff don’t tend to stay very long and retention’s very low, it will almost be the champion, kind of a right-hand AI man.
KA: It’s that co-pilot thing, isn’t it? Obviously, Microsoft has adopted that phrase, “co-pilot for your Microsoft Office 360,” but it’s an obvious choice of word. It’s what everybody needs, isn’t it? Everyone needs a co-pilot to get their kids booked into all their clubs after school and organize schedules. To pay the utility providers and constantly remember to insure the car and manage emails – it’s far too complicated. I’m reasonably tech savvy, and yet I find the complexity of modern life difficult. I can’t imagine how tricky life admin must be these days for the less techy among us.
PG: They are tapping into a real need. You laid out the forces driving this particular trend, and they seemed really on point. I did see a few virtual assistant-type things made to look like people. I don’t know whether it’s a generational thing or just an individual thing, but I find them a little creepy.
KA: I think that’s true. I think everyone finds them a bit creepy. It’s the uncanny valley.
RW: Kate was talking about [AI personal assistant] Pi, which has a natural voice that really sounds like a real person. And I think the computer imagery and CGI is now coming so fast, when you can merge the two. It’s actually not going to be creepy then. I think we’re so close to that. Even in robotics, the hyper-real robotics are now incredible, but they just need to merge that with the AI as well. When they all come together, it’s going to be so much more acceptable. At the moment, it’s slightly off, because you look at them and you think, oh, that’s just really weird. The one I was talking to you about, she looks slightly angry, so they’ve got quite good functionality, but maybe she doesn’t look quite right just yet, but I don’t think we are far off.
PG: I can believe that, because it’s already advanced this far.
KA: When you think of how fast ChatGPT has changed, the speed of change now is what’s exciting.
PG: Speaking of that, I saw a vendor in the [NRF] Innovation Lab, and they had a solution where they collected all of this information on a product, all of this background information so that people, if they put it in a search engine, they could pull up all of this information. In order to create this collection of information, the vendor used AI, but they said they specifically fed the AI engine things like users’ manuals and other accurate information to eliminate, as they called it, “hallucination.” That’s one of the things people who are against AI are afraid of, that it makes up things because its information is based on all of the internet, which has plenty of misinformation.
KA: It does totally make sense, doesn’t it? If only 1% of the world’s data is actually on the internet, because most of it’s behind firewalls, it does totally make sense that each business would take a LaMDA 2-type model, a large-language model that is off the shelf, and then configure it to their own vertical data sources, which are more accurate, more appropriate to their particular subject. I think that’s obviously going to be the way to go.
PG: So, the third trend was creating immersive experiences in real life. In stores, have you seen any good examples of grocers who are able to do that?
KA: I think there was a division between pre-COVID, when that was pretty hot with lots of higher-end supermarkets, which were moving in the direction of, “We’re a cooking show and we’re a bar, and we’re a restaurant at night.” They were everything the community wanted: tuition and kids’ creches and yoga and all. Then COVID sort of shut that all down, didn’t it? Then it was all e-commerce. And I think now we’re on a sort of slow trajectory to start to find a new idea of what does experiential grocery shopping look like. And I don’t think it’s yet back to where we were before.
Honestly, I think there’s so much great technology you can implement into the grocery retailers. I wonder what’s holding retailers back from doing some major innovations. It’s not so much in store, though, is it? It’s the quick delivery. That’s a major innovation convenience. But yet you go to most supermarkets these days, and it’s still the same.
[Read more: “Are You a Retail Tech Innovator?”]