AI in the smartphone changes phase: from gadget to personal assistant

For years, artificial intelligence on smartphones has been used above all to correct photos by cutting out ‘unwelcome’ subjects, putting filters and recognizing the environment to optimize the result. Small automationsalmost like gadgets.

Today, according to the main manufacturers, the very nature of the mobile experience is changing: from an “extra” function to an integrated assistant, capable of acting on many daily micro-activities, with an impact that is measured more in time saved than in special effects.

From gadget to personal assistant

Giorgia Bulgarella, marketing director of Motorola talks about the cultural passage: at the beginning, AI was “perceived but also communicated as a gadget”, useful for differentiating a not very distinct market. Now, however, “it is evolving” and is “more perceived as an assistant, but also used as an assistant”.

In the Motorola case, he cites the growing use of “Moto AI” as a sign that people are “starting to interact.” And look further: the biggest leap it will be towards “agentic artificial intelligence”, in which the ecosystem counts more than the single device: many interconnected devices, “the individual and the agent at the centre” capable of proactively learning habits and needs.

Samsung pushes on the same concept in a practical way. Nicolò Bellorini, vice president of the South Korean company in Italy, insists that AI must really be useful: “The important thing is to provide services that are useful, that are useful, not simply that help to generate headlines in the newspapers.” And it explains the real effect: an increase in productivity that arises “from the sum of very small increases” on daily operations, from browsing long documents to photo retouching, video improvement and “natural language” translations, without forcing the user to learn new apps or procedures.

On-device or cloud? The answer is hybrid

The big debate is where AI “lives”: on the device or in the cloud?Huawei, with Andreas Zimmer (Head of Product for Europe), defends a historical philosophy: “We tried to put as much as possible on the device”, i.e. AI “on the edge”. And it indicates two advantages: not only privacy, but also speed (“practically zero latency”). The cloud remains necessary when “information from external sources” or broad knowledge is needed, but “whenever possible, the solution should be” on-device.

Samsung converges on the hybrid model but puts it at the center data governance: “there will be both. The important thing is to give the possibility to choose, that is, to be able to decide where the data should be”. And it connects this promise to security and transparency tools: the Knox platform and privacy controls in the interface, to decide “at any time” whether the data remains on the device or can be shared in the cloud and to manage and revoke app permissions. It’s not just the most powerful AI that wins, but the one that gives the user and companies the most confidence in where the data ends up.

AI as an “engine” of new forms of interaction, even on leaflets

Where the A becomes more visible is in the new form factors. Samsung, speaking of foldables, describes devices that “redefine the mobile experience” as tools for productivity and creativity, thanks to advanced functions. And he cites concrete examples, such asmore advanced multitasking and new ways of interacting.

Honor, for its part, gave demonstrations of “physical AI” at the MWC. Pier Giorgio Furcas, commercial director of the Chinese company, introduces an interesting idea: the camera is no longer “passive” but “active”, capable of “exploring the surrounding world” and communicating with the context, approaching the concept of assistant “tout court”.

The “invisible” constraint: chips and memory

The other side of AI is hardware: more AI means more need for computation and memory. Furcas says it directly: “for artificial intelligence there are two elements: a chipset and memory”. And he warns that growing demand “is driving up costs”, especially on the memory side, which is why Honor is aiming for a “very strong partnership” with suppliers with a strategy: if costs increase, features must be added to justify the value.

Motorola also links the evolution of AI to pressure on the supply chain. Artificial intelligence is now “embedded” even in the medium range, so from a cost issue for the consumer it has transformed into a problem of components and planning, because when demand runs faster than forecasts “there is a lack of raw materials”.

Dimon hu

Paid services and subscriptions: a real prospect or still premature?

At this point the problem for companies is how continue to earn even after selling the smartphone. Will we pay subscriptions for “smart” features? Talks at the Mobile World Congress in Barcelona suggest caution. Bellorini insists on control and integration in the experience. Furcas observes that today the user tends to pay more willingly for entertainment than for telephone features, and that the possible monetization role could be played more by operators with integrated offers than by brands with aggressive paywalls.

Honor, with Dimon, shifts the focus to overall value and the ecosystem: “Bringing value to the consumer not only through the smartphone, but through the entire ecosystem”, with a strategy that includes PCs, tablets, wearables and AIoT.

The promise is that AI will become less of a demonstration and more of an everyday thing. A function that permeates the smartphone and makes it more personal, faster and – if the manufacturers keep their word – more controllable.