Human-centric AI: digital innovation starts with people
In the debate on artificial intelligence and digital transformation, the concept of ‘human in the loop’ is no longer sufficient. And for one simple (though not obvious) reason: merely maintaining a human presence in automated processes risks reducing people’s role to that of mere final oversight of decisions taken elsewhere. The challenge facing businesses and public administrations today runs much deeper, and calls on every organisation to design technologies that are built around users’ needs, their behaviour and the social context in which they will be used. We must therefore embrace a vision of ‘human-centric AI’ in which innovation, inclusion and regulatory responsibility go hand in hand – a vision that concerns not only the adoption of artificial intelligence but, more generally, the way in which digital services and operational models are conceived, developed and implemented.
This new perspective on AI, in line with a European rather than a national approach, is clearly reflected in the evolution of e-commerce and conversational interfaces. According to the first European study on agent-based commerce published by Sopra Steria, in fact, more than half of European consumers (55% to be precise) are already familiar with AI-based shopping agents – tools capable not only of assisting users but also of directly managing certain stages of the purchasing process. Over the next ten years, more than €310 billion worth of online transactions in Europe could be facilitated by these intelligent tools, whilst between €62 billion and €77.5 billion could be managed using these technologies within the next three years. The growth of these models will be driven not only by the level of technological maturity but also by trust, which will become the real deciding factor.
In this regard, the study highlights that 74 per cent of European consumers still identify significant friction points in the online shopping experience (ranging from the difficulty of comparing offers and products to after-sales service), and consequently show an interest in tools capable of simplifying the customer journey. Conversely, 94% of users state that they wish to retain ultimate control over purchasing decisions, highlighting a trend that is by no means insignificant: in short, the adoption of AI cannot be based solely on automation and efficiency, but must incorporate transparency, comprehensibility (including the ease of cancelling purchases), data protection and the ability to adapt to people’s expectations. It is precisely within this context that the ‘human-centric’ approach to digital transformation takes shape, based on technologies genuinely designed around people, usage contexts and organisational sustainability.
Designing with real needs in mind
Inclusive design of digital services is one aspect of this evolution, and takes shape in the development of solutions (which, of course, comply with regulatory requirements or technical accessibility standards) that can be used simply and effectively by as many people as possible. Underpinning this model are research and co-design methodologies that involve real users from the very earliest stages (through interviews, behavioural observation, workshops, ongoing testing and participatory creativity) with the aim of addressing not only permanent disabilities but also temporary vulnerabilities, cultural differences and various forms of digital literacy gaps. The issue of accessibility, moreover, is becoming a strategic priority, and the entry into force of the European Accessibility Act is accelerating the updating of digital platforms in this regard – not only for compliance purposes, but as a means of improving the relationship with the user-consumer in terms of experience and personalisation.
When we look at the public sector, this approach becomes crucial for ensuring equitable access to services available online or via apps. The redesign of certain platforms operated by the Lombardy Region (including the Unified Employment Information System (SIUL), the IDO job-matching system and the SIUF training portal), managed by Sopra Steria in collaboration with ARIA, was driven by the aim of overcoming process fragmentation and the limitations of legacy architectures through an inclusive co-design model that involved accredited bodies, service providers and citizens/end-users right from the earliest stages of the project, Sopra Steria, in particular, combined multidisciplinary teams and ‘agile’ methodologies to foster an iterative prototype design model, which made it possible to simplify digital pathways, improve service accessibility and increase the level of interoperability and integration between the platforms. Furthermore, the IDO project has begun integrating artificial intelligence to support self-service activities and improve the matching of labour supply and demand.
AI and processes: innovation driven by human experience
The same principle applies to the introduction of AI into companies’ day-to-day operations. The proliferation of generative models and autonomous agents is prompting many organisations to carry out rapid trials, often driven more by competitive urgency than by a defined and structured strategy. However, it is now well known that the effectiveness of an artificial intelligence project depends less and less on the technology itself and increasingly on the ability to integrate it into processes that are consistent with people’s needs. For this reason, the most mature organisations are adopting approaches that begin with an analysis of operational contexts, the identification of priority use cases and an assessment of organisational and cultural impacts. In this sense, the study and analysis of behaviour becomes a decisive factor, and Sopra Steria’s research on agentic commerce highlights how consumers are willing to delegate to AI activities perceived as repetitive or technical (such as searching for products, comparing offers and even managing certain stages of the customer journey) but remain much more cautious when sensitive issues such as health, finance or money management come into play. What does this trend among end users tell us? That designing conversational interfaces or intelligent agents cannot be limited to the algorithmic component alone but must necessarily take into account expectations, levels of trust, perceptions of risk and the nature of the relationship with the brand.
It is in this context that the ‘human-centric AI’ paradigm establishes itself and distinguishes itself from a purely technological approach, and the case of Unieuro – Italy’s leading consumer electronics and household appliances chain in terms of number of shops and geographical coverage – serves as a prime example. The redesign of the e-commerce platform, also carried out by Sopra Steria in this instance, arose from the need to meet the requirements introduced by the European Accessibility Act and to update a user experience that was no longer in line with the evolution of the digital customer journey. The project addressed both this aspect (leading to a simplification of workflows and a reorganisation of content to ensure greater clarity and readability) and the technological architecture (with the adoption of a new platform based on cloud-native microservices) and was developed using design thinking methodologies and user research, with the aim of creating simpler, more inclusive and accessible navigation paths.
Change is built together with people
The organisational dimension represents the third key element of the ‘human-centric’ philosophy. In many digital projects, change management still comes into play ‘downstream’, when the technology has already been introduced and it is necessary to manage resistance or difficulties with adoption. In models that place people at the centre of the digital solution design process, however, their involvement is an integral part of the design workflow right from the start. What benefits does this new approach bring? The most significant advantage lies in the ability to engage internal and external stakeholders from the very earliest stages, to support teams in redefining processes, and to work on organisational culture as well as technological tools.
The issue becomes even more relevant in the context of artificial intelligence, because the transformation no longer concerns only operational activities but also the relationship between technological autonomy, responsibility and trust. ‘Human-centric AI’ is therefore a concrete alternative to a vision focused exclusively on the speed of technology adoption – a model in which the value of innovation goes beyond the ability to automate processes and is measured by the potential to build technologies that are more accessible, understandable and sustainable for people, organisations and society.
