AI And Shopping Habits Collide In Italy’s Retail Shift

On January 26, 2026, two stories collided at the heart of Italy’s consumer landscape and the global debate over artificial intelligence: the evolving habits of Italian shoppers in the non-food sector and the growing controversy surrounding AI-generated information sources like Elon Musk’s Grokipedia. Both tales, though distinct, share a common thread—how technology is reshaping the choices people make and the trust they place in the information at their fingertips.

According to the latest GS1 Italy Non Food Observatory report, presented during Marca by BolognaFiere & ADM, Italy’s non-food sector achieved a turnover of 111 billion euros in 2024, marking a modest yet steady increase of 0.6% compared to the previous year. With 433,000 products available, the sector remains a substantial part of the Italian economy. But beneath these numbers, the way Italians shop—and the tools they use to decide what to buy—is undergoing a quiet revolution.

Physical retail is seeing a notable rationalization. Instead of countless small outlets, there’s a shift toward commercial hubs, concentrating a diverse range of offerings in single locations. The aim? To make shopping more convenient for consumers who, especially when it comes to durable goods with higher price tags, are increasingly inclined to do their homework before making a purchase. One out of every three Italian consumers now seeks information about non-food products on the internet before heading to the checkout.

But here’s where things get interesting: among those who search online, half are turning to AI-powered search engines for answers, while 40% still prefer the wisdom of human reviews. The type of product matters too. AI is most frequently consulted for consumer electronics, followed by clothing, footwear, underwear, accessories, and sports articles. Unsurprisingly, it’s the younger generations who are more likely to trust the judgment of artificial intelligence, though, across the board, people still lean toward human interaction when making their final decisions.

"Non-food products require a large amount of information, so it’s important to meet this need and build a relationship of trust with consumers. In this process, artificial intelligence can be a useful tool, provided the data entered is truthful, precise, and transparent. Otherwise, the outputs produced will be unclear, if not outright wrong, undermining the trust the brand has built," explained Samanta Correale, Senior Business Intelligence Manager at GS1 Italy, as reported by GS1 Italy.

The Observatory’s findings also shed light on the channels Italians use to make their purchases. Specialized stores maintain a strong presence, accounting for 36% of sales by value, while large specialized surfaces (GSS) follow closely behind with 32%. E-commerce, though not present in every sector, averages 12% of sales where available. Some categories are more digital than others: for edutainment, a whopping 58% of sales happen online, far outpacing the channel’s overall average. Meanwhile, GSS dominate in clothing and sports articles, covering 50% and 57% of sales value, respectively. Even traditional hypermarkets and supermarkets have a role, handling 28% of stationery sales and 17% of household goods.

Across all 13 non-food segments monitored by GS1 Italy, growth has been the story of the past five years. Four major trends are shaping this trajectory. First, there’s a growing focus on wellbeing, with consumers seeking products that support physical and mental health, stress management, and improvements in home and outdoor experiences. Second, the "kids-adult" phenomenon is blurring generational lines, as adults increasingly purchase and collect items once considered the domain of children—think toys, comic books, and video games. Third, sustainability is coming to the fore, with a "buy for life" philosophy encouraging second-hand purchases, refurbished goods, and repair services over disposable consumption. And finally, the demand for "everything, always, and immediately" is pushing retailers to seamlessly integrate online and offline experiences.

Speaking to the sustainability trend, Davide Rossi, General Director of Aires Italia (which represents major electronics retailers like Euronics, Expert, GRE, MediaWorld, and Unieuro), predicted a dramatic shift: "Sales of repaired or refurbished products are estimated to increase from the current 3% to 30% in 10 years. This is an opportunity for brands to assume a leadership role in the regenerated products market, but there’s a need to fill the gap in precise regulations that set clear limits and govern relationships between retailers and manufacturers."

Yet, as Italian consumers become more reliant on digital tools—especially AI—to guide their choices, the reliability of those tools is being called into question on the world stage. On the very same day as the GS1 Italy report’s release, The Guardian broke a story that sent ripples through the tech community: OpenAI’s GPT-5.2 language model, widely used to power search engines and digital assistants, had been caught drawing information from Grokipedia, Elon Musk’s AI-generated encyclopedia.

Launched in October 2025 as a counterweight to what Musk described as Wikipedia’s "left-leaning" tendencies, Grokipedia is built almost entirely by Musk’s AI model Grok, without the open community editing or independent quality control that characterize traditional encyclopedias. Critics have long warned that Grokipedia’s content is not only factually unreliable but also slanted toward right-wing narratives. The Guardian’s investigation found that GPT-5.2 accessed Grokipedia content nine times across more than a dozen queries, including sensitive topics like Iranian organizational structures and biographical details about historian Sir Richard Evans—some of which were found to be "plainly false." In some instances, GPT-5.2 even prioritized Grokipedia over more accurate, established sources.

The implications are troubling. If AI models use content generated by other AIs as their foundation, errors and biases can be amplified in a feedback loop—a phenomenon summed up by the old adage, "Garbage in, garbage out." The Guardian also highlighted the risk of "LLM grooming," where actors intentionally seed the internet with AI-generated material to shape the knowledge bases of future AI systems. This process can gradually shift the information ecosystem from human-verified facts to synthetic content, making it harder for users—and even for AI itself—to distinguish what’s real from what’s not.

OpenAI maintains that GPT-5.2 is designed to draw from a broad array of publicly available sources and that safety filters are in place to block harmful or unreliable content. However, as The Guardian’s reporting shows, these filters don’t always catch AI-generated sources like Grokipedia, raising tough questions about the ability of today’s most advanced AI systems to reliably separate fact from fiction.

As Italy’s non-food sector adapts to new consumer demands and technological realities, and as the world grapples with the consequences of AI-driven information, the need for transparency, accuracy, and trust has never been more urgent. Whether it’s choosing a new washing machine or searching for the truth about world events, the stakes are high—and the answers, it seems, are only as reliable as the sources we trust.