AI and data science in 2026: the trends everyone’s talking about

When it comes to AI in retail, 2026 is shaping up to be a pivotal year. The experiments are getting more ambitious – and the expectations higher. But beneath the noise, a more interesting picture is emerging. Discussions around what’s new and exciting are giving way to questions about what really drives value. The hype hasn’t gone away, but it’s being tempered by a more grounded, pragmatic approach to AI.

To help cut through the noise, Chief Data Science Officer at dunnhumby, Sandra Stanley, is back with her AI and data science trends for the year ahead – three areas she expects will dominate industry conversations, and three that will determine whether those discussions lead anywhere useful.

Agentic Customer Experiences

2026 will be the year that shoppers begin interacting with AI agents that behave more like concierges than chatbots. Not full autonomy, we’re not quite ready for AI that spends our money unassisted, but more meaningful than a conversational search engine.

If it’s going to succeed, the future of agentic has to be about helping people get things done across the entirety of their shopping journey – not just at checkout. Agentic AI will need to help shoppers plan meals, build lists, compare options, discover new products, and make decisions with less effort.

We’re already seeing progress here. Google’s new Shopping Assistant, for instance, blends product understanding, availability data, and recommendation logic, making it feel more like an actual helper rather than a glorified search engine.

Retailers who provide flexible APIs, trustworthy data and agent-compatible journeys will become preferred partners in these experiences. Those who rely on “walled garden” thinking will lose relevance.

Could 2026 be the year this concept moves from hype cycle to practical application?

Retail media (fuelled by AI) will transform every surface into smart inventory

Retail media has been on a rapid upward trajectory for years, and 2026 could add some extra fuel to the fire as AI starts to reshape how it works day-to-day – both operationally and creatively.

Contrary to a lot of the conversation that’s currently playing out, AI’s primary role here won’t be to invent wild new creative ideas. Instead, it’ll be about elevating existing brand-safe assets and adapting them intelligently at scale. With the right data and asset library, AI can blend brand content with retailer guidelines, dynamically adjust creative, and personalise content based on local context or audience.

What does this mean in practice? That hyper-localisation suddenly becomes simple, enabling advertisers to tailor their message, imagery, or emphasis depending on store location, region, or even time of day.

As exciting as this sounds, though, there’s some nuance at play here too: good AI-powered media will still need great underlying assets and metadata to drive it. Perhaps most importantly of all, the creative executions need to be up to scratch; a Coke bottle needs to always look like a Coke bottle, no matter how many versions appear across screens. Creative variety can be powerful, but it’ll only work if the inputs are precise.

Spatial AI sets out on the long road from concept to reality

The third trend we’ll hear a lot about in 2026 is the rise of spatial AI: the label for a group of technologies that give AI access to physical stores through sensors, cameras, shelf monitors, computer vision, and employee-worn devices. In other words, we’ll be able to give AI a sense of location, space, and place.

Let’s be crystal clear: spatial AI won’t reach full maturity until around 2028 at the earliest and likely much later. But 2026 will be the year it becomes part of industry conversations. You’ll be expected to at least have a view on it, if not a fully-fledged strategy. There are good reasons to engage here, too: around the world, retailers are already using spatial to detect shelf gaps, monitor product movement, identify stock issues, and direct colleagues more intelligently.

Interestingly, some of the most impressive advances are coming from startups, including companies using small cameras on employee badges to build live digital models of a store. The flip side of applications like that, of course, concerns data: as well as investing in the front-end tech, retailers will also need to be able to process the new kinds of data that result.

Now let’s look at the other three trends – less visible, perhaps, but vital for turning those ambitions into reality.

Data quality becomes paramount

We’ve discussed the exciting prospect of agentic experiences and posited the potential of AI-powered media. But here’s the thing: none of that works without strong data foundations. High-quality taxonomies, product graphs, availability feeds, and structured metadata might not be making headlines – but they’re still the backbone of every AI trend we’ll hear about next year.

Take Google’s Shopping Graph, a real-time dataset of products, prices, and availability. That rich pool of information is what’s powering its new shopping assistant, constraining model outputs to verified information only. That principle applies across retail: without reliable foundational data, agentic tools will misfire, personalisation will be inconsistent, and on-brand retail media will be impossible at scale.

Trust could become a flashpoint without serious focus

In 2026, trust won’t hinge only on privacy and cyber security, although the events of this year have shown they remain essential foundations. What’s changing is that trust is expanding beyond those basics. Responsibility, transparency and fairness are becoming the new battleground. Customers don’t care how AI works until something goes wrong.

We now expect AI systems to explain themselves, behave consistently, avoid biased outcomes and comply with the EU AI Act and other global standards. However, complicating matters is the probabilistic nature of generative AI: these systems produce what’s likely, not what’s certain. Without grounding in accurate, trusted, domain-specific data, they can get things wrong, confidently and at scale. And in a world where mistakes spread instantly through social channels, the reputational stakes are high.

Regulation will always lag innovation. Culture won’t. Retailers with strong governance, the right mindset and responsible AI practices will be the ones able to innovate safely and boldly, while others pull back.

The meaning of value will continue to evolve

After the excitement comes the reality. Organisations that have been investing in AI for the past few years are starting to ask sharper questions about what that spend has brought them. What will the actual benefits be? How will they be measured? Will they really be realised? And with around 75-95% of AI proofs-of-concept failing to return value (depending on who you listen to), those questions are entirely fair.

Expect 2026 to be a year of rationalisation. As contracts come up for renewal, organisations will start to define use cases more deliberately, distinguish productivity gains from cost savings, and put real emphasis on the value created by AI. The focus will shift from “we need to invest” to “we need to understand what this investment will deliver.”

Stay tuned for more predictions from across the world of grocery retail soon.