by Faisal Kawoosa
On-device AI is the new thing in smartphones. One after another we are witnessing smartphone OEMs introducing their AI smartphone. So, is it worth buying?
AI or artificial intelligence is a game-changing technology or a layer of interface between the environment including users and the device. This capability lets the device be more contextual, perform advanced computing and extract great value out of the data that is on the device. Before we evaluate if is it worth buying an AI smartphone, let us first try to understand an overview of on-device AI.
On Device AI
So far, the galaxy of devices with which we interact with the Internet, or any other network has had limited computing capabilities. They couldn’t intelligently compute or in other words, think and compute. They would simply process the data, which by any means isn’t that simple too. Now with on-device AI, these devices including smartphones have enhanced edge computing. This means these devices can not only process but also intelligently process data and extract greater value out of the data making it contextually significant. In the case of smartphones with AI-powered chipsets both from Qualcomm and MediaTek, the smartphones now have dedicated Neural processors or NPU, in some cases, it is also called APU (AI processing unit).
In the old days architecture of processors, we primarily had a CPU, which can be seen as a generic purpose processor capable of doing all tasks but not specialised ones. One could think of a CPU as a ‘jack of all trades’! Then came the era of having dedicated GPUs or graphics processing units which were able to process images and videos to support the heavy lifting of data and use cases like gaming, etc., evolved further only to become more realistic.
Now with on-device AI, we have a dedicated NPU/APU in chipsets or processors. These are specifically designed to process not millions but billions and trillions of operations per second. This is why we measure the efficiency and performance of NPU/APU in TOPS (Trillion Operations per Second). The NPUs or APUs are specific processors made to process LLMs or Large Language Models – the language, grammar and dictionary of AI.
On-device AI smartphones have NPUs or APUs as part of the SoC in them along with the supporting LLM. In most smartphones today, Google’s Gemini is being used as the underlying LLM though there is considerable customisation on top of it by the respective OEMs. So, we can think of Google Gemini as the operating system equivalent of AI in these smartphones.
The on-device AI smartphones primarily leverage the processor capabilities of the device itself to compute the data on the edge and get the desired results. Of course, it needs to talk to the cloud to access some parts of LLM like photo libraries to achieve the results. So, we shouldn’t be thinking that on-device AI smartphones will work without internet connectivity. But yes, we may easily conclude that the capabilities of early-day supercomputers are now in the pocket!
AI Smartphones
Now that we have got the perspective about on-device AI and how it works, let us now evaluate whether it is worth owning it. Well, to be honest, at present the AI-powered features and functions may not have a very compelling use case that we would recommend a ‘must buy’ for these. Be it enhancing the pictures or doing very creative edits with them, using AI to help create and summarise content, or for that matter helping in cross-cultural communications, these are not the use cases that make use cases compelling enough to make a buy decision.
Let’s for example have a look at some of the recently announced AI smartphones in India to understand this better. We had Motorola announcing its new flip smartphone Moto Razr 50 Ultra, Samsung introducing the new generation (6) of Z fold and flip smartphones and Oppo announcing the Reno12 series. In all these phones AI is powered by Google’s Gemini, and they use one of Qualcomm’s Snapdragon or MediaTek’sDimensity chipset that has an AI processor. The use cases showcased and experienced are no less than magic and add considerably to the creativity, productivity and convenience coefficients of users. For example, one can do unimaginable level of photo editing without any professional skills. Similarly, one can create great content or enhance existing to serve professional tasks.
But all these use cases have a workaround or jugaad as we love to know it in India. So, they are not fundamentally unique or bring novelty. It either would require advanced skills otherwise or one would need to spend some good time across different platforms and machines to achieve the same. With AI smartphones, it’s now just easy!
To buy or not?
AI smartphone doesn’t open any new unique use cases that make it a ‘must buy’. Unlike some transitory moments of the past, like from featurephone to smartphone, from smartphone to camera smartphone, from 3G to 4G, etc., there were tangible additions that were unique to a smartphone. This compelled everyone to eventually upgrade to one and even get prepared to pay a premium to own one. That is not the case with AI smartphones yet. It is too early I would say. However, the good thing is that smartphone OEMs are looking at AI as an essential addition that has to percolate across consumer segments. So, they have to make a beginning which they have already made in the luxe (above Rs 50,000) segment and now even in the premium (Rs 25,000 – 50,000) segment.
This means users who can afford to or are considering buying smartphones in these segments can look for AI smartphones to buy. I would recommend that they buy an AI smartphone as it wouldn’t cost them disproportionately more. The devices announced don’t have a premium because of AI. Samsung announced Fold&Flip 6 just Rs 10,000 more than the previous year’s prices and with exchange and bank offers can own one at less than the launch price of Fold & Flip 5 respectively. Similarly, Motorola is also offering Moto Razr 50 Ultra for Rs 10,000 more than the previous edition, but it has hard bundled buds with it. Oppo has on the contrary reduced the price of Reno 12 pro to its predecessor by Rs 10,000 or more depending on how you buy it.
Conclusion
From a utility point of view, it may be too early to consider an AI smartphone. There may be some features that may excite only once, or the excitement might fizzle out in just a few days. But from the routine upgrade and replacement point of view, if as a user you are considering buying a premium or luxe smartphone, then go for it. You will not only be getting early access to some of the interesting and creative features powered by AI making you a distinct owner of such features, but also join the early AI wave to become part of tech enthusiasts and trendsetters who can also contribute to making AI more useful through feedback and in-use review of AI smartphones. Even if you won’t proactively share any feedback, the usage patterns and trends will anyways pour into the smartphone OEM analytics engine which can make a lot out of it to make future experiences better. So, if you are in the zone, let us AI!
(The author is a recognised industry analyst specialising in gadgets, AI, 5G, IOT and Auto Tech. Founder of Techarc, he offers insights to leading Technology brands and consults startups and mid-sized organisations for digital transformation. Ideas are personal.)