AI and Emotions: What Will the Customer Experience Look Like in 2050?

In 2023, the global market for AI-based solutions applied to customer journeys was already worth $53.2 billion (1). Today, artificial intelligence is reshaping this market, while emotions play a key role in purchasing decisions. Between automation and hypercustomisation, how will customer relationships evolve in the coming decades? We asked Arne De Keyser, professor at EDHEC, and Alexis Rollin (EDHEC class of 1993), founder of the French agency RnD.

Chatbots are still one of the most popular applications of AI. How will these tools evolve in the future?

Arne de Keyser: In recent years, chatbots have made significant progress. We are Moving from decision-tree models — with preset replies to a set of predetermined questions — to chatbots driven by large language models (LLMs) that respond more precisely and naturally to users’ queries. That said, these ‘new’ chatbots will only be effective if companies have sufficient data to train the underlying models. A generic LLM-bot will be able to respond fluently, but it will not provide the answer a user is looking for. Another emerging innovation is voice integration (2). After the initial interest some years ago with assistants like Alexa and Siri — which didn’t quite take off — several companies are now coming out with much more promising voice-enabled solutions. Some of these voice bots have become remarkably advanced: It’s nearly impossible to tell they aren’t human.

Alexis Rollin: At RnD, for instance, we’re rolling out new websites later this year that add a microphone icon to the traditional menu bar, letting visitors simply talk to their phone or on-page assistant. I believe that in the future, brands will need to fully embrace the voice-first experience. But to be clear, the EU AI Act will ban any system that pretends to be human, effective August 2026. We expect machines to save us time — not to impersonate us.

Brands have always tried to build emotional connections with customers to encourage loyalty. In one of your studies (3), Arne, you differentiate between “mechanical AI,” “thinking AI” and “feeling AI.” Do these systems affect consumers differently?

AK: Mechanical AI is primarily used to automate routine tasks (like using an ATM). These gradually Evolved into “thinking AI-systems,” which mimic certain human cognitive functions in order to make decisions. Netflix’s recommendations are a good example here. Over 80% of what is being watched on the platform is driven by algorithms. More recently, we have seen an interest in “feeling AI,” which is designed to detect and respond to human emotions. It’s important to stress that AI doesn’t feel emotions itself, yet some research shows that we’re capable of forming emotional connections with these AI-bots. Our study (3), conducted in hotel and restaurant settings, revealed that advanced AI systems may create stronger and richer customer experiences.

Can AI really detect consumers’ emotions? How far can it go in personalising the customer experience?

AK: AI can indeed be used to analyse human emotions — for example, if a customer calls customer service and raises their voice, AI will likely flag this as “anger.” However, analysing emotions is not Always clear-cut; feelings can be expressed very subtly and may be difficult to detect. But let’s be honest: Not all humans are equally good at detecting others’ emotions, either!

AR: On the emotional front, AI is making rapid progress. In the realm of text, the sentiment analysis used to understand brand perception on social media has Advanced dramatically — AIs can now detect subtleties like irony. For visual input, there’s still work to be done, but software that can analyse body language is starting to emerge. What I find interesting is that AI is now making the old marketing dream of the 2000s possible. Remember one-to-one marketing? Back then, it was more a concept than a reality. Today, AI is tipping the scales towards hyperpersonalisation. Soon, two users landing  on the same webpage won’t see just a 20% difference in content, but 50% or even 60%. Armed with data and behavioural insights, companies will tailor each interaction to maximise conversion.

By over-analysing data, aren’t brands at risk of losing their Identity or originality? Isn’t there a danger of uniformity?

AK: In marketing, we’re seeing a triple shift: We’re spending more time on social media, we’re watching more videos, and producing those videos is getting cheaper. So logically, we’re flooded with content. To stand out and be favoured by chatbot algorithms, brands will need to break through the noise — and I firmly believe authenticity is the key to this: crafting a unique tone to build trust and drive conversion. 

Generative AI can help save precious time that teams can then dedicate to creative tasks that define the brand’s distinctiveness.

In the future, consumers will likely rely on conversational agents to make purchases. For a brand to remain visible, it’ll need to appeal to algorithms. That’s a radical change in approach.

AR: We’ve entered the era of “agentic AI.” Today, the goal for brands is no longer just to be found (as is the case with traditional SEO, or search-engine optimisation), but to be understood. Companies will need to create structured content — short, concise answers, diagrams, infographics, etc. — that AI can explore and present to the end user via a digital (shopping) assistant.

AK: The real winners in this space will be those who manage to stand out through personal assistants, which are poised to become the go-between for consumers and countless businesses. Just last week, OpenAI announced the international launch (everywhere except in Europe, where they need to make adjustments due stricter regulations) of an extended memory feature, which allows AI to remember all user interactions and provide even more refined responses — a major step toward next-level personalisation.

AR: But there’s a risk: what we call the “Eliza effect” — our tendency to attribute genuine understanding and personality to a program that only simulâtes conversation — which can dull our critical thinking. That critical thinking can and must be taught and cultivated. That’s also how we’ll learn to deal with “LLM poisoning” (manipulating the data used in AI training, which could lead to ethical issues, bias or fake news). Every innovation has its downside: In this case, that includes disinformation (among other things), and it’s up to us to put the right guardrails in place.

So ultimately, is the key to the future of commerce a successful collaboration between humans and AI?

AR: Absolutely, and that’s what I always tell the teams I train: It’s a choreography between humans and AI. Personally, I strongly believe in phygital expériences and complex customer journeys, especially when the service calls for it. For a vending machine, human interaction adds no value. But in a hardware store, a chatbot might give me a starting point, which prompts me to seek out a salesperson for deeper advice and empathy. I also think we need to move past the fantasy of a 100% automated world. I often tell clients to proceed step by step: Reaching 30-40% automation already frees up time to focus on creativity and business uniqueness.

AK: I agree. We won’t erase the human factor. Today, around 80% of retail sales still happen offline, where humans play an important role. And even if we could automate everything, would we want to? That’s a really important societal question. How far do we want to go? To me, as humans are inherently social beings, we’ll always need some human connection in business. In fact, it could even become a point of differentiation. In the Netherlands, for example, the supermarket chain Jumbo opened a dedicated checkout line for customers who prefer interacting with a cashier. Those customers know it may take longer — but they choose the human touch and conversation. Rather than aiming for total automation, we should think about how humans and AI can work together. AI is improving constantly — but we need time to adapt to it and learn how to interact with it.

References

(1) AI market projected to hit $4.8 trillion by 2033, emerging as dominant frontier technology, UN Trade and Development (UNCTAD), 7 avril 2025

(2) Henkens B, Schultz CD, De Keyser A, Mahr D (2025;), "The sound of progress: AI voice agents in service". Journal of Service Management, Vol. ahead-of-print No. ahead-of-print - https://doi.org/10.1108/JOSM-06-2025-0269

(3) Larivière, B., Verleye, K., De Keyser, A., Koerten, K., & Schmidt, A. L. (2024). The Service Robot Customer Experience (SR-CX): A Matter of AI Intelligences and Customer Service Goals. Journal of Service Research, 28(1), 35-56 - https://doi.org/10.1177/10946705241296051 (Original work published 2025)

(4) Yenee Kim & Richard G. McFarland, 2024. "Are you looking for something specific or just looking around? Adaptive selling on the basis of customer shopping goals in retail sales," Journal of the Academy of Marketing Science, Springer, vol. 52(6), pages 1780-1804, November - https://ideas.repec.org/a/spr/joamsc/v52y2024i6d10.1007_s11747-024-01015-y.html

See also Adaptive selling: how personalised sales strategies can restore the competitive edge of physical retail (2025), EDHEC Vox