How Microsoft Sees Generative AI Revolutionizing Retail
At the Shoptalk Europe retail conference, Microsoft's Shish Shridhar tells WWD why the technology is here to stay despite concerns about ethics and data protection.
BARCELONA, Spain — With generative AI at the heart of every conversation, Microsoft's stand was among the buzziest on the opening day of the Shoptalk Europe retail conference in Barcelona on Tuesday.
Attendees thronged around executives including Shish Shridhar, worldwide lead for retail with Microsoft for Startups. Shridhar has more than 15 years of experience working with top retailers to help solve business challenges using
AI, machine learning and emerging technologies.
At a time when the potential for generative AI to revolutionize working habits is generating equal parts horror and fascination, he was predictably upbeat about the impact of the technology.
Through its Microsoft for Startups Founders Hub, the tech giant helps business-to-business start-ups to scale with in-house services including its Azure cloud computing platform. Through its partners, it also offers fundraising support and access to GPT-3, the neural network machine learning model developed by OpenAI that uses deep learning to produce human-like text.
In an interview with WWD, Shridhar talked about some of the start-ups he's working with in fields as diverse as fashion design, sorting systems and store design, and why — despite concerns about ethics and data protection — he believes generative AI is here to stay.
WWD: So the buzzword at this conference is obviously AI. What is Microsoft presenting? Shish Shridhar:
We have a bunch of AI-based solutions, though the big one is generative AI. It's the one that's taken all of the attention.
One of them is a company called Fashable, and they're using generative AI for creating new designs. So they've been in a space where designers are trying to create new designs, they're looking at lots of signals: what is trending, what colors, what designs, and all of that, and the AI model actually looks at all the data signals, trying to figure out what is happening out there for different demographics, looking at social media, and then creating these new designs for brands.
The other ones are, I would say, more operational. For example, there's a company here called Unbox Robotics. Unbox Robotics is more about using robots and AI to do sorting for shipping, so they'll kind of do the efficient sorting, but it's all completely automated.
Yet another one is for planograms in the stores. So there is a company called Omnistream that is here with us, that is using AI for efficiently setting up the layout for a store using all kinds of data.
And yet another one is a company called Pecan. What they do is, they address the challenge of the shortage of data scientists, so they're using no-code bots, or enabling this as prebuilt templates, so you can do things like determining churn, predicting sales.
WWD: One attendee pointed out that procurement processes in companies are quite slow, but with these types of solutions, retailers do need to move faster, because there will be a firstmover advantage. Do you think that’s the right way to look at it? S.S.:
Absolutely, and a lot of the AI capabilities are actually enabling them to move a lot faster as well, get the insights from all the processes and reduce the process overload that many of them have. One interesting company that I'm working with is called dSilo. They'll take all of these documents and contracts from the suppliers and vendors and all of that, and then analyze everything so you don't have to read all the contracts. It will optimize all the contracts and look at, are there bottlenecks? Are there overlaps with some of the work you're doing with different vendors? And it will consolidate everything. So absolutely, yes, and AI is enabling, making all those things more efficient, cutting down the processes in a big way as well.
WWD: Do you think ethical issues are going to be a significant barrier to retailers adopting some of these technologies? S.S.:
There will be some barriers when they're trying to understand the implications, because right now, I think a lot of them are holding off, making sure they understand the implications of the AI models and whether they're putting in compliance and guardrails around those things. So if there is a slowness in adoption, it was mainly in retailers saying, we understand the value they bring, but at the same time, we want to do the due diligence and ensuring that there are guardrails, and that we understand it. So there is concern, for sure.
WWD: In terms of data ownership, if a retailer has a chatbot powered by ChatGPT, how do they ensure that their customers’ data remains private and contained? S.S.:
So, by design, the ChatGPT models can be set up where you can have your private databases chained to it with none of that data leaking over. But then at the same time, it is important that internally there are experts doing the audits to make sure that the systems are set up in a way that that private data isn't leaking over, and is being used in a way that the personal and confidential data is not being put directly into ChatGPT. So I think the audits and controls need to be put in place to ensure that there is a separation. But can it be done? Absolutely. And this would be similar to any other system where, when you put a system in place, are there security challenges? Are there compliance challenges? Yes. And ordinarily, every retailer would be doing compliance and audit checks to make sure that it's done the right way, and that same thing can be done with ChatGPT.
WWD: How fast would you expect retailers globally to be adopting generative AI technologies? Is this something they need to do before the end of this year? S.S.:
At the rate that it's moving, and the benefits they can get from it, I think they can actually start using it now. I'm seeing the interest level as well. Every conversation I'm having with friends is around generative AI. It is almost like they are at a race trying to figure out what we can do with it. With every other technology, it's been the other way around, where they're saying, I will identify the problem, then I'll figure out which tool. But with generative AI, it has been going the other way around, where they're having sessions trying to figure out what are the use cases that are the best. So there is a desire to move very quickly on their part as well, and it has proven itself that there are very quick wins you can get with it. And as a result, I think by this year, there's going to be some significant use cases. I'm seeing it with start-ups adopting and enhancing some of the solutions that they have.
That's been amazing. They're not creating new products, but they're enhancing what they have, they are embedding the generative AI into their tools.
Once the technology becomes invisible, that's when it's really useful. And in the case of generative AI, right now there's so much attention on it and so much hype on it, and there are a lot of useful use cases being created. At some point, we're going to forget about the generative AI; it's just going to be an element within a solution.
WWD: Yes, because right now the conversation is about whether it’s going to wipe out humanity. It’s not really about sorting boxes. S.S.:
That happened with AI. A couple of years ago, everyone was hyping AI.
But then at some point, it became table stakes. It was part of everything. You didn't have to say AI; AI was in there. And I think with generative AI, we are seeing that same thing happen where it is going to be a component.
We're going to stop talking about generative AI. That's when it's made it.