Thanks to Meta AI, there's no avoiding artificial intelligence (AI) even if you want to. On Monday, the AI-powered smart assistant rolled out across Meta apps like Instagram, Facebook and WhatsApp in India, one of Meta's most important markets, ushering in the era of artificial intelligence anywhere, anytime. This means billions of users will be exposed to the technology for the first time, and it's hard to avoid. AI will come to the forefront in the interfaces of the applications users use every day.
Meta's AI is marketed as a conversational chat window that can ask questions and generate images. Its actions and behaviors are very similar to OpenAI's ChatGPT, Microsoft's Copilot, and Google's Gemini. OpenAI's ChatGPT is popular, but it's not yet part of people's daily experience. Though it has a large user base, very few people actually use ChatGPT for everyday tasks. This is what makes Meta AI unique: it's embedded deep within the application interfaces you use most often on your smartphone throughout the day, making AI technology an integral part of it.
The AI chatbot is seamlessly integrated into the search and messaging features across Meta’s family of apps, and processing happens in the cloud. If you don’t see Meta AI yet, check back later. The best way to identify Meta AI is by looking at its logo. It has blue and pink rings that glow from time to time. On Facebook, when you tap on the search icon at the top, you’ll notice that the search bar has been replaced with one that says “Ask Meta AI anything.” On Instagram, WhatsApp, and Messenger, you’ll notice that Meta AI has taken over the search bar and appears as a separate chat. To interact with Meta AI, start typing your questions or random things. Meta AI can do a lot of things, like animate images, request news summaries, search things like reels, and more. This is essentially the same idea as ChatGPT, which is built inside WhatsApp or Instagram.
AI is Everywhere
Meta's aggressive efforts to bring artificial intelligence technology to its most popular apps shows how big tech companies are desperate to bring AI technology into their digital lives. The social media giant started slowly by adding generative AI to its apps as a beta feature in certain markets to test user reactions to a technology that is still in its early stages and whose widespread impact is unclear. However, given that Mark Zuckerberg's Meta is looking to expand its AI technology, the company now wants to extend AI to its most popular apps as soon as possible. The strategy is expected to put it ahead of competitors by leveraging AI in a unique way and making the technology mainstream through its large user base.
But Meta isn't the only company betting on artificial intelligence and integrating it meaningfully into the most used apps and consumer devices. Last week, Apple unveiled its implementation of AI, Apple Intelligence. Apple Intelligence is not a generative AI app or a chatbot like ChatGPT. Rather, it's an integrated layer of artificial intelligence built into Apple's upcoming operating system. Apple Intelligence's AI features include an improved Siri that can get summaries of web pages, help you compose emails, and even let you create your own emojis at will.
Microsoft has also integrated its Copilot AI chatbot into its Windows laptops and core applications. Google is not lagging behind in the AI race either, accelerating the development of AI models and introducing them into products like search, docs, Gmail, and Android phones. Well-funded startups like OpenAI, Anthropic, and Perplexity are looking to integrate cutting-edge AI models into more apps and services.
Meta's installed base is seen as an advantage
Meta's aim to bring AI technology to billions of people at scale also comes from the fact that users don't need to invest in or pay for new devices to use Meta AI. At least for now, this gives Meta an edge over other technology companies. Take Apple Intelligence for example. Technically, Apple isn't charging users for its AI features, but the problem is that Apple Intelligence runs on-device and requires a lot of processing power, meaning it's limited to Apple's latest devices with specific Apple Silicon chips. This approach forces users to upgrade to new devices to use Apple Intelligence, which Apple clarifies it's not doing on purpose.
Microsoft is doing the same. The company has various Copilot features in everything from Office to Teams, but its big consumer push is integrating Copilot into its Windows and Surface experiences. That means to get the full Copilot experience, you need to buy the premium Copilot Plus PCs, which are powered by Qualcomm's ARM chips, released last week. These devices have powerful neural processors and can deliver a full AI experience with on-device AI. Meanwhile, smartphone makers looking to bring AI features to their phones will need to integrate Google AI to get access to the latest technology. But integrating Google AI features comes at the cost of upgrading devices to use the latest generative AI features. Others, like Amazon, are considering charging for their AI-powered Alexa voice assistant.
Meta's “open source” AI model approach is different
Meta's focus on the potential of generative AI and unlocking its true potential to its advantage also comes from how the company is building the AI models that power its AI capabilities. Meta is one of the few technology companies that has “open-sourced” its AI models, meaning anyone can incorporate its AI technology into their apps and services for free. This gives Meta a major advantage over OpenAI and other major AI companies that are not ready to open-source their own AI models.
Meta describes its latest large-scale language model, LLaMa 3, which underpins its Meta AI assistant tool, as “the best open model on par with the best proprietary models available today.”
Large-scale language models and foundational models play a key role in AI. These are complex software that make predictions and decisions using algorithms trained on huge amounts of data. Technology companies need to continuously train and develop these AI models to improve their capabilities. However, training AI doesn't come cheap. It's estimated that OpenAI's latest model, GPT-4, cost about $100 million to train, several times more than GPT-3.
However, open-sourcing Meta is not new. Technology companies have used open-source technologies to catch up or get ahead of their competitors. Google did the same when it open-sourced its Android mobile operating system to compete with Apple's iPhone. This single move gave Google a lead in the market and captured a large portion of the smartphone segment.
Privacy concerns remain an issue
But privacy watchdogs and activists have raised concerns about how Meta uses data to train its AI services. The company's privacy record has been called into question, and Meta's desire to make the buzzy technology ubiquitous by placing AI chatbots across its apps has raised questions about what the social media giant does with people's information. But Meta says it complies with privacy laws and respects people's privacy while training its generative AI models.
“We did not train these models using people's private posts, nor did we use the content of private messages with friends and family to train our AI,” Mehta said in a blog post last year.
Meta has been using public posts to train its AI models in the U.S. and other markets where Meta AI operates. However, user backlash forced the company to suspend training its AI models with data from users in the European Union and the U.K. For now, Meta is delaying the release of Meta AI in Europe due to privacy and regulatory barriers.
“We are disappointed that our lead regulator, the Irish Data Protection Commission, on behalf of European DPAs, has asked us to delay training our large language models (LLMs) with public content shared by adults on Facebook and Instagram, particularly as we have taken regulators' feedback into account and notified European Data Protection Authorities (DPAs) since March,” Mehta wrote in a blog post.
The lack of trust in Meta and how the company uses public data to train its AI models may be the biggest hurdle for the largest social media company in making AI ubiquitous. While privacy laws are less stringent in India and the US, data protection regulations in Europe and the UK prioritize consumer privacy. Europe's GDPR and DMA regulations are tough on big tech companies, especially Meta, which requires vast amounts of data to train its AI systems. This is true for all tech companies looking to deploy AI capabilities on their devices and apps.
Trust in tech companies is already at an all-time low due to the way they have been run for years. But new laws and regulations are forcing companies to buckle and change their business models and products. Apple is reportedly delaying the launch of its Apple Intelligence-generated AI tool in Europe, citing EU laws. This is despite the company's assurances when announcing Apple Intelligence at its recent developer conference that most AI processes are handled on-device by Apple's processors, but some will have to go through cloud servers to take advantage of larger and more complex language models. Apple says it has built secure private servers that don't store any information, but has not said whether users will know their information is leaving their device or if they can turn it off. OpenAI's partnership with ChatGPT has raised privacy concerns, but Apple has made it clear that the ChatGPT maker will not store any personal data.
