Meta and Google, like OpenAI and Apple, persuade new users as AI space evolves
Google’s Gemini is proficient in English and 9 Indian languages. Meta AI, for now, only understands English.
Days after Google finally made artificial intelligence (AI) assistant Gemini available to users in India, it is Meta’s turn to unlock Meta AI within WhatsApp, Instagram, Facebook, and Messenger. If you are curious about the timing of Google and Meta’s moves, it has all to do with acquiring active users, as Microsoft builds its Copilot integration across smartphones and newer Windows PCs, alongside Apple’s upcoming Apple Intelligence suite for iPhone, iPad, and Mac users.
Meta AI’s strengths include search the web, text to image tools, get contextual inputs such as organising projects or trip itineraries in group chats, create stickers, learn more about a post on the feed and can be called upon for suggestions or answers to queries.
“The benefit of the way that we’ve rolled out Meta AI is that it’s available in all of our apps, regardless of the platform. You can access it in WhatsApp, whether it’s on Android or iOS. You get it on the web, and we really want to be everywhere that you are. Right now, our apps provide both the biggest distribution and most availability to the end user,” Ryan Cairns, who leads engineering for AI at Meta, tells HT.
India is Meta’s biggest market and makes for a logical port of call. WhatsApp, according to research firm Statista, has 535.8 million users in India, with Brazil in a distant second (118.5 million). Facebook clocks 378.05 million active users here, significantly ahead of US (193.8 million) and Indonesia (119.05 million). For Instagram, India’s contribution of 362.9 million active users, surpasses US (169.65 million) and Brazil (134.6 million).
India holds similar importance for Google. Android, according to research firm StatCounter, has a 94.98% market share in the country – far ahead of a global average of around 69% share, and more than markets including US (around 44%) and UK (around 49%). Gemini, now available on the Play Store, can replace Google Assistant. A Gemini Pro 1.5 model is available to AI Premium (that’s ₹1,950 per month) subscribers.
“We’re unlocking new features in Gemini Advanced such as new data analysis capabilities and file uploads, and also launching the ability to chat with Gemini in Google Messages, starting in English,” says Amar Subramanya, vice president of engineering, Gemini Experiences, Google.
Localisation is important for AI assistants, for audience in India. Google’s Gemini is proficient in English and 9 Indian languages. Meta AI, for now, only understands English. OpenAI’s latest GPT-4o can work with 12 Indian languages.
For Meta, this is work in progress, but they aren’t revealing specifics. Cairns does refer to the fact that Meta AI’s Llama 3 model is updated every couple of weeks. This means adding languages, alongside fixing hallucinations, accuracy and safety, wouldn’t take much time.
This should tie in well with a push for responsible AI. This week, OpenAI’s Chief Technology Officer Mira Murati wrote in a post, “realizing the potential of AI is not guaranteed. This means putting strong safeguards in place, reducing harmful biases, and proactively dealing with potential negative effects.”
In May, Google updated their best practices for responsible use of open models such as Gemma. Earlier this month, Microsoft suffered an embarrassment of having to recall an AI feature called Recall, designed to take screenshots of everything users do on newer generation Copilot+ PCs.
Preparing for an AI future
OpenAI’s position of strength is unique. It provides an all-important model foundation for Microsoft’s Copilot and holds an important role within Apple Intelligence’s jigsaw for iOS 18, iPadOS 18 and macOS Sequoia. A smarter Siri, and more apps, will rely on integration of their latest GPT-4o models.
Some of Apple Intelligence’s capabilities will include rewriting emails and drafting responses such as RSVPs, changing tone of text you’ve written, recording phone calls with transcriptions, fix handwriting in notes, picking actionable pointers from web pages, image generations or Genmojis as a language of conversation. Apple’s foundation models are trained on AXLearn framework, an open-source project released in 2023.
While Apple Intelligence is slated for release around October, Apple’s making a point to not refer to it as ‘artificial intelligence’. Apple CEO Tim Cook describes this as a “personal intelligence system”. Once it is activated in entirety, it’ll have scale of integration not seen till now on Android, Windows PCs, or any other platform.
Apple’s non-compromising approach to privacy is why it developed Private Cloud Compute servers, for tasks that cannot be done on-device.
“Your request to our servers is anonymized, so your IP address is masked. Then it talks to a server that has no permanent storage, cannot log, but most importantly is running software where the image is publicized for security researchers to audit,” explained Craig Federighi, Senior Vice President for Software Engineering at Apple, to HT.
OpenAI has also detailed privacy measures in place for queries generated from Apple devices such as when accessing ChatGPT within Siri and Writing Tools. User requests are not stored by OpenAI, and users’ IP addresses will be obscured, they say.
