A spark that illuminated the light: For AI phones, ecosystem aligns its pieces

For ‘AI phones’, a moniker we’ve collectively willed into existence, there can always be a deliberation about what the exact spark was. There have been moments, that seemed pivotal. OpenAI’s evolving models? Chatbots as smartphone apps? Microsoft’s persistence with AI that since got rebranded from Bing to Copilot? Google’s Bard assistant, followed by Gemini models? Or their Pixel 8 Pro becoming the first true ‘AI phone’ with the Gemini Mini model overlaying AI functionality never seen before in a smartphone?

These may be developments providing the momentum, but not everyone is aboard the AI ship just yet. OnePlus’ latest flagship, the OnePlus 12 announced a few days ago, distinctly stays away from any conversation about AI. For now. Yet, distinct is a new-found spirit of collaboration that’s pushed the case for AI phones. Think about it. Samsung and Google, who’re otherwise competitors in a tough smartphone market, are instead curating AI experiences in sync. The ecosystem is aligning, as chipmakers such as Qualcomm, app developers and AI companies collaborate. Does it underline a new era we’re embarking on? Questions remain, some unanswered.

The most pertinent perhaps being, are we moving to a time when we’ll have to pay subscription for using AI features on smartphones? A hint comes from Samsung’s India website listings of the Galaxy S24 phones, which reads, “Galaxy AI features will be provided for free until the end of 2025 on supported Samsung Galaxy devices. Different terms may apply for AI features provided by third parties.”

There are secondary questions around it too – will the entire AI suite, called Galaxy AI, be behind a subscription paywall or users can select individual features to sign up for? Is Google implementing the subscription, or Samsung? If a user opts out of paying for AI features come 2025, what happens to the data collected for AI tasks till then? What’ll happen to the remaining functionality of any app once AI features are disabled?

HT reached out to Samsung India and Google, but haven’t received any response with clarifications to the planned Galaxy AI subscription roadmap.

“The Galaxy S24 represents a strategic bet on Samsung’s part to leverage AI for purpose-driven and powerful user experiences. This aligns with a broader industry trend in 2024, where smartphone OEMs will increasingly aim to unlock new possibilities leveraging AI” says Prabhu Ram, Head- Industry Intelligence Group (IIG) at Cyber Media Research (CMR).

Samsung’s own Exynos chips, which also power the company’s Galaxy S24 and Galaxy S24+ phones sold in some countries, take away some market from Qualcomm. But the latter is only too happy to customise their flagship Snapdragon 8 Gen 3 chip for the Samsung Galaxy S24 Ultra (it is called the Snapdragon 8 Gen 3 for Galaxy), with focus on AI that’s been closely developed with Google.

“Snapdragon 8 Gen 3 for Galaxy instils its advanced AI capabilities in the Galaxy S24 series, to enable new experiences with AI features to empower users’ everyday life. It also fuels advanced professional quality camera, gaming experiences and ultra-fast connectivity including Wi-Fi 71, plus offers one of the most reliable authentication solutions available with our Qualcomm 3D Sonic Gen 2 technology,” says Chris Patrick, senior vice president and general manager of mobile handset, Qualcomm Technologies, Inc.

This isn’t the first time. Last year’s Galaxy S23 phones also took advantage of a customised chip solution from Qualcomm. The chipmaker is pursuing performance boosts for generative AI tasks on phones that run their Snapdragon chips. The Snapdragon 8 Gen 3 mobile platform is the first of its kind to support gen AI models with up to 10 billion parameters, on-device. That’s alongside AI pipeline improvements for unlocking functionality for photo editing and audio.

Google’s made its new contextual Circle to Search feature, an exclusive for Samsung’s Galaxy S24 phones and on their phones, Pixel 8 and Pixel 8 Pro. Google’s Gemini Pro model will be built into the Samsung’s Notes, Voice Recorder and Keyboard apps. Imagen 2, which is Google’s advanced text-to-image diffusion technology, unlocks photo editing capabilities using Generative Edit, within Samsung’s Gallery app. The Gemini Mini model will be part of Google Messages app, for the on-device processing of Magic Compose suggestions. There will also be Photomoji, to create a new emoji from a user’s photos using generative AI.

Amidst the excitement about AI on Samsung’s new phones, there’s a word of caution too. “While the premium smartphone segment growth provides a favourable backdrop, success of Samsung’s new phones will hinge on their ability to seamlessly integrate and evolve AI features beyond past iterations. Samsung must deliver transformative capabilities that set the Galaxy S24 apart from competition and offer a compelling value proposition for consumers seeking a fundamentally distinct upgrade,” adds CMR’s Ram.

In December, Google had rolled out the Gemini Mini model underlier exclusively (at the time) for its Pixel 8 Pro phone running the new Tensor G3 chip, enabling on-device generative AI functionality including summarising audio files in the Recorder app, smart replies in Gboard and a gamut of camera functions such as removing smudges from document scans and one-tap photo unblurring. That made it the first AI phone, in the truest sense.

But there’s more beneath the top layer.

Conventional logic pits the Google Pixel phones and Samsung’s Galaxy S phones, as competition in the flagship Android space. Who cares anymore? AI might have just played its part in bringing two competitors to the same table. Google has good reason – they know Samsung ships the most Android phones globally and there cannot be a better platform to get a significant demographic of users interacting regularly with their AI models. That’s more data too, for AI to improve

“With our latest advances in AI, we have the opportunity to enhance what billions of people already love about Android — whether it’s accessing information quickly, connecting with people that you care about, capturing the perfect image or expressing your personal style,” says Hiroshi Lockheimer, Senior Vice President for Platforms & Ecosystems, at Google.

The point of no return for AI phones?

What Samsung and Google, with Qualcomm providing the crucial compute power, have done is ensured the approach to smartphones will fundamentally change. It is now time for phone makers, and AI companies, to walk the talk we’ve heard in the last year. That is, customising generative AI tools for smartphones and the mobile-first use cases.

Microsoft, for one, is on a relentless pursuit. The tech giant has not only rebranded Bing AI chat to Copilot and thus giving it cross platform uniformity, there are feature-loaded apps for Android phones and Apple iPhones that are now available. Here’s what the Copilot app can do – an AI chatbot, drafting emails, summarising chunks of text and a text to image generator. This means Microsoft is now at parity in competition, as more users attempt AI adoption, with OpenAI and Perplexity’s apps.

The company has also rolled out a Copilot Pro subscription (that’s ₹1,999 per month, over a Microsoft 365 subscription; though Copilot remains free – it can get quite complicated) across its suite of apps and services (this will have some relevance on phones too, including the Copilot app) with OpenAI’s GPT-4 Turbo model (the free Copilot use gets you access to GPT-4) and customised Copilots.

Yusuf Mehdi, executive vice president and consumer chief marketing officer at Microsoft, also talked about a new Copilot milestone. “We have reached another milestone in this mission with more than 5 billion chats and more than 5 billion images to date. As Copilot continues to earn preference and usage, we’re receiving valuable feedback on how to improve,” he said, in a statement.

Apple is yet to reveal any elaborate plan for AI integration within the iPhone, iPad and Mac computing devices, but it is likely that a revamped Siri assistant will be at the core of the transition expected with iOS 18 later this year. Things may already be in motion. Surely, Apple has a trick up its sleeve.

It must not be forgotten, the path for AI implementation and on-device compute was first unveiled by Apple with the iPhone X and its A11 Bionic chip as far back as in 2017. Then, this silicon introduced a Neural Engine and on-board machine learning capabilities. That method still remains relevant, and every year, there’s a push towards faster AI processing.

Android phone makers joined the race much later but as Samsung and Google prove, have since moved ahead with user-interfacing AI implementation.

Apple’s iOS 17.2.1 update from late last year, has enabled a Live Voicemail functionality – it unlocks an on-device voicemail on an iPhone, which is invoked if you do not manually accept an incoming call, and shows a live AI generated transcription of a message a caller may be saving on the voicemail at the time, on the screen. The voice to text functionality also negates the need for having to listen to a voice message.

It may not be too distant into the future when more phone makers, including the likes of Xiaomi with their revamped HyperOS software, OnePlus, Vivo and Nothing, make substantial moves with AI tools that may bring a dollop of exclusivity to their next line of phones. They’ll have to, since the market may be moving in that direction.

According to CyberMedia Research, the super-premium smartphone segment (that’s price points between ₹50,000 and ₹1,00,000), grew 63% year-on-year in 2023. Within that, phones priced more than ₹1,00,000 recorded a staggering 148% growth in sales. They expect the Samsung Galaxy S24 series to corner 8% share of the super-premium segment in 2024.

In the latest Smartphone 360 Service report, Counterpoint Research expects ‘GenAI’ phones to collect about 8% of the total smartphone market, up from 4% in 2023, and progress towards a share of 40% by the year 2027. It is no surprise that chipmakers are ready for the transition, as the latest chips from Qualcomm and MediaTek, as well as Google’s own Tensor G3 and Apple’s A17 Pro, are geared for faster neural compute and AI processing.

We’re now well and truly, in a different era. Sooner than we may have expected.