Release of next-gen AI products for merchants via Google Cloud

This is Google’s plan: to bring some creative AI into the retail sector. (Or at least to give it a go.).

New-generation artificial intelligence tools were launched today by Google Cloud in conjunction with the National Retailer Association’s annual conference in New York City. These technologies will assist retailers in streamlining back-office processes and personalizing online shopping experiences.

Since TechCrunch was not granted access to test the new tools before their announcement this morning, the writer cannot comment on whether or not they work as promised. (At some point in the first quarter, they will debut.) Whatever else you want to say about Google’s relentless pursuit of next-gen AI clients, the deluge of announcements proves it.

Similar to a brand-specific ChatGPT, Conversational Commerce Solution is one of the latest offerings from Google Cloud that allows businesses to install agents powered by AI on their websites and mobile applications. The assistants engage in natural-language interactions with customers, providing personalized product recommendations that cater to their unique tastes.

The use of chatbots with brand names on them is nothing new. Agents powered by “sophisticated” models like PaLM, according to Google, may be fine-tuned and personalized using data from merchants’ own sources, such as catalogs and websites.

This new catalog and content enrichment toolkit from Google Cloud works in tandem with the Conversational Commerce Solution. It uses generational AI models like PaLM and Imagen to automatically create product descriptions, metadata, classification suggestions, and more from a single product photo. Retailers may also utilize the toolkit to create new product pictures from current ones or to develop AI-generated product shots based on product descriptions.

A few months ago, eBay introduced a feature that uses artificial intelligence to convert product images into descriptions. However, sellers quickly started complaining about the feature’s effectiveness, citing instances of misleading, repetitious, and even false content.

If Google has done anything to address concerns about these hallucinations, I wanted to know from Amy Eschliman, who is Google Cloud’s managing director of retail. Although she could not provide any concrete examples, she did emphasize that human review is integral to the catalog and content enrichment processes and that Google is “continuously improving” its tools.

When lives are on the line, I hope there is human review. After all, it’s not completely out of the question that a store may get into trouble with customers or face false advertising claims due to an inaccurate AI-generated picture or description in a catalog.

Enterprise use cases may benefit from human-in-the-loop by ensuring high quality, reducing bias-related risk, improving trust and transparency, improving and continuously training the model, and complying with regulatory and company policy, according to Eschliman.

Similarly, Google unveiled a Distributed Cloud Edge device tailored to the retail industry today. This controlled, self-contained hardware package aims to “reduce IT costs and resource investments” associated with retail using AI. (While Google has long provided Distributed Cloud Edge as a service, the company is now aiming its marketing efforts squarely toward shops.) Designed to power clients’ next-gen AI applications, the edge cluster is available in sizes ranging from one server to many servers. Google claims it can fit inside establishments such as petrol stations, convenience stores, fast casual restaurants, and grocery stores.

Even if a store’s physical location loses internet access for a few days, Google Distributed Cloud Edge will allow them to continue working smoothly, according to Eschliman. A small cluster of nodes maintained by Google Cloud is now available to retailers and can be easily set up in almost any shop. Now that these technologies and software are completely controlled, stores may run their current software with distributed AI, allowing mission-critical activities to function at all times.

Google has promised to provide Q1 price and availability details.

After getting the rundown on everything, my concern was this: are stores truly eager to use Gen AI?

I suppose so. Certainly the retail behemoths.

Yesterday, Walmart revealed its heavy investment in generation AI search, which would let customers search using particular use cases (e.g., “unicorn-themed toddler birthday party”) and better grasp the context of searches. At the same time, Amazon has been using generation AI to compile review summaries, assist sellers with writing product descriptions and picture captions, and improve the sizing experience for customers.

Google found that a large majority of retail decision makers (81%) see the need to implement general artificial intelligence (gen AI) into their company immediately, with 72 percent saying they are prepared to do so. This generation of AI is most commonly used for customer service automation, marketing, generating product descriptions, assisting with creativity, facilitating conversational commerce, and enhancing the knowledge and support provided by store associates.

When I think about the recent shaky launches of generation AI in retail (e.g., Amazon’s review summaries that overstate negative comments), I’m not sure the sector will be quick to embrace generation AI in droves, whether it’s from Google Cloud or another source. I guess we’ll just have to see what happens.