It seems we can’t escape it. Every new presentation of a smartphone is now accompanied by an inevitable mantra: “It has artificial intelligence”. Manufacturers repeat it to us like a modern incantation, promising that this time yes, that the next device we buy will be radically smarter and will transform the way we use it.
But you can’t help but stick with one feeling of déjà vu. Are we really facing a revolution, or is it simply the fashionable slogan that has replaced “more megapixels” or “bigger screen”?
Faced with this bombardment, it is logical to ask ourselves an uncomfortable question: Does it really make sense to be seduced and change your phone just because it “has artificial intelligence”? And, beyond the marketing slogans, it is worth asking ourselves whether we really need a new terminal to enjoy these capabilities or whether, on the contrary, We are faced with an empty label that hides a more complex reality. Let’s break it down.
Samsung, Google, Apple… they all want us to buy their new AI-powered mobile phones, but does it make sense?
Mobile phones with artificial intelligence They are not just a marketing strategyas it seems. There is a fundamental reason why manufacturers incorporate this concept into their latest generation smartphones. To understand this, it is necessary to distinguish between the two main architectures that dominate the market: AI on the device and AI in the cloud.
When you use an application like ChatGPT or access Perplexity from its website, you use artificial intelligence in the cloud, compatible with virtually any mobile phone on the market, since the only requirement is an Internet connection to communicate with remote servers. On the contrary, On-device AI requires manufacturers to integrate neural processing units (NPUs) specialize in their chips, such as the Apple Neural Engine or Google Tensor.
This local architecture offers significant advantages in terms of privacysince the data is processed directly on the device without the need to transmit it. Furthermore, it guarantees operation without an Internet connection. However, the performance of the artificial intelligence on the device it is limited by hardware capacity and can consume more battery in intensive tasks. Likewise, model updates are generally slower, as they depend on new versions of the operating system or applications.
Instead, Artificial intelligence in the cloud, used by services like ChatGPT, Gemini or Grok, exploits the almost unlimited power of remote servers. Its main advantage lies in its enormous computing power, high scalability, and the ability to instantly implement model updates. On the other hand, it is completely dependent on a stable internet connection and poses more privacy risks as it requires sending external data for processing.
The problem is that local AI has not proven to be a better option than cloud AI
Now, there’s a crucial detail that manufacturers usually don’t highlight: Even phones with dedicated AI chips rely heavily on cloud servers for their more advanced features. That is, you buy premium hardware, but the “magic” still happens outside of your device. The reason? The enormous power required for certain tasks would not only dangerously overheat the phone and drain the battery, but it would also it wouldn’t get the same results.
This is not a theory, but something that we have seen confirmed in two emblematic examples:
- Samsung’s “slip”.: The company announced the Galaxy S24 as the first “AI phone,” but months later moved nearly all of the Galaxy’s AI features to the previous model, the S23. This clearly demonstrated that previous generation hardware was already perfectly capable of running them, as most AI functions are performed in the cloud.
- The democratization of Google– Tools like Google Photo Magic Editor, which were previously exclusive to Pixel, are now available for any Android or iPhone for free or with a very affordable subscription.
It’s clear then the hardware of “AI phones” in real-world use doesn’t make a big difference. Manufacturers are turning to AI in the cloud more than you might think to achieve quality results in their “local AI capabilities” without depending on mobile performance.
The AI you use really doesn’t need a new chip
Think about AI tools you use mostlike ChatGPT or Gemini. I am, The most powerful and popular ones work mostly in the cloud. For complex tasks, like programming or document analysis, you get superior performance from remote servers, not the phone’s chip.
This brings us to a key revelation: The AI functions that most of us use (generating text, making summaries or writing code) do not depend on having the latest processorbut rather an Internet connection and sometimes a subscription. The AI chip (NPU) inside the phone provides negligible value for these tasks.
There is no need to buy an expensive mobile phone: artificial intelligence is now within everyone’s reach
For the average user, the next generation NPU It doesn’t justify the expense of a new cell phone. You should only consider it if you have very specific and stringent privacy needs that prevent you from using the cloud (also knowing that local AI delivers worse results).
But for 95% of users, The AI we actually use (chatbot and photo editing) works perfectly on the phone you already own. Don’t get carried away by marketing. If you need specific AI on your current smartphone, simply install it via an app. Here is a list of 22 best free artificial intelligences that will be of great help to you.
