Google wants to inject a little generative AI into retail. Or at least we try.
In conjunction with the National Retail Association’s annual conference in New York City, Google Cloud today unveiled a new generation of artificial intelligence products designed to help retailers personalize online shopping experiences and streamline their back-office operations.
As for whether it works as advertised, this writer can’t say — TechCrunch wasn’t given the opportunity to test the new tools before unveiling them this morning. (It’s scheduled to launch sometime in the first quarter.) But a slew of announcements show, if nothing else, how aggressively Google is trying to woo AI customers.
One of Google Cloud’s new products, Conversational Commerce Solution, allows retailers to embed AI-powered agents on their websites and mobile apps — a bit like a brand’s ChatGPT. Agents have conversations with shoppers in natural language, offering personalized product suggestions based on shoppers’ individual preferences.
Branded chatbots are not new. But Google says “sophisticated” models like PaLM power agents, which can be fine-tuned and customized using retailers’ own data (such as catalogs and websites).
Complementing the conversational commerce solution is Google Cloud’s new catalog and content enrichment toolset, which taps into generic AI models including the aforementioned PaLM and Imagen to automatically generate product descriptions, metadata, classification suggestions, and more from just one product image. The toolkit also allows retailers to create new product images from existing products, or use product descriptions as the basis for AI-generated product images.
Now, when eBay launched a similar AI-powered product image-to-description conversion capability a few months ago, it didn’t take long before sellers started complaining about its performance — citing misleading, needlessly repetitive, and in some cases not entirely truthful text. .
I asked Amy Eshleman, managing director of Google Cloud’s retail division, what steps, if any, Google has taken to address concerns about such hallucinations. It did not indicate specific measures but stressed that Google is working to “continuously improve” its tools and that human review is an essential part of its catalog and content enrichment workflow.
I sure wish there was a human review when the stakes are high. It’s not unlikely, after all, that a misleading AI-generated image or description in a product catalog could put a retailer in hot water with shoppers — or on the receiving end of false advertising claims.
“Human-in-the-loop is a best practice that helps enterprise use cases ensure high quality, mitigate risks associated with bias, improve trust and transparency, and continually improve and train the model, while complying with regulatory and business policy.” Eshleman said.
In a related announcement today, Google unveiled Distributed Cloud Edge for retail, a self-managed family of devices to “reduce IT costs and resource investments” around AI for the retail generation. (Google has long offered distributed Cloud Edge as a service, but is now targeting retailers more directly.) Available in a range of sizes from single-server to multi-server configurations, Google says the Edge suite is designed to fit stores from the convenience of marketplaces and terminals. Fuel to fast-casual restaurants and grocery stores – powering general AI applications for customers.
“With…the control plane running locally, Google Distributed Cloud Edge provides retailers with non-stop operations even when their sites are taken offline for short periods of time (days),” Eshleman said. “Retailers now have access to a small set of nodes managed by Google Cloud that can be easily installed in almost any store. With this fully managed hardware and software, retailers can now run existing software with distributed AI to enable mission-critical operations in Store at all times.
Google says pricing and availability information will be released in the first quarter.
My question after being briefed on all of this was, frankly, are retailers really clamoring for AI generation?
maybe. At least the retail giants.
Walmart announced yesterday that it is investing heavily in general AI search to better understand the context of queries and allow shoppers to search by specific use cases (e.g. “toddler’s unicorn-themed birthday party”). Meanwhile, Amazon is leveraging artificial general intelligence to summarize customer reviews, help sellers write product descriptions and image captions, and better enable buyers to find clothes that fit their sizes.
In a survey conducted by Google, Google says that 81% of retail decision makers feel a “urgent need” to adopt AGI into their business while 72% feel ready to deploy AGI technology today – specifically in the areas of customer service automation and marketing support Product description creation, creative assistance, conversational commerce, knowledge storage and support.
But given some of the difficult launches of AGI in retail recently (see: Amazon review summaries that exaggerate negative reactions), I can’t say I’m convinced that the retail industry will rush to adopt AGI en masse — from Google Cloud or any other provider. I suppose we’ll have to wait and see.