Meta Connect 2023: Everything Brand Marketers Need To Know
Meta offers crucial updates regarding its AI and metaverse strategies, signaling another attempt at consumer attention land-grabbing
On Wednesday, Meta hosted its annual Connect conference and unveiled a slew of interesting new products and features. Unlike the previous Connect events, originally launched as Oculus Connect, typically focused on Meta’s ambition in building its metaverse platform and mixed reality headsets. This year, however, generative AI was undoubtedly the star of the show, as Meta seeks to leverage the massive scale of its social and messaging platforms to bring consumer-facing AI features to mainstream users, as it hopes to boost engagement (and ad dollars) with AI-backed conversational interfaces.
AI-related announcements not only dominated the stage time during the keynote, it even percolated into Meta’s hardware announcements as well. While the Oculus Quest headset got the third-generation upgrade that many were anticipating, it was the other hardware product, the Meta Ray-Ban smart glasses, whose user experience received a major boost from Meta’s AI integrations.
Here are the three biggest announcements from the event, how they impact the current landscape of consumer-facing generative AI and mixed reality, and why brand marketers should pay attention.
Meta Goes AI: a New Era for Conversational UI
Among all the things that Meta announced on Wednesday, Meta AI, a “general use” chatbot that Meta created based on its LlaMA-2 model, stands out as a foundational piece of Meta’s strategy to productize generative AI across all of its platforms.
Back in mid-July, Meta introduced LLaMA-2, the next generation of its large language model (LLM), in an extended partnership with Microsoft. Like many LLMs, LlaMA-2 is a text-only AI. But with Meta AI, a voice mode has been added so that users can actually talk with the AI chatbot, which removes the friction of written communications and makes the user experience feel more natural and fluid. Meta AI will be available across all Meta platforms: from Facebook Messenger to Instagram DM, anywhere users are using Meta’s services to chat with someone, the Meta AI can be summoned to answer questions and provide information. The Microsoft partnership persisted as Meta AI will access up-to-date information via Bing search.
Moreover, a new image generation model named Emu, which is basically Meta’s answer to Dall-E, is being integrated into Meta AI, making it a true multi-modal chatbot. Not only you’d be able to chat with Meta AI, you can also ask it to create images for you. To that end, Meta is using the new Emu image generator to create fun stickers in Facebook Messenger and WhatsApp, and power new photo-editing tools coming soon to Instagram.
If you’re getting a case of déjà vu from this, you’re not alone. Back when it was still called Facebook, the company made a big push for chatbot integration circa 2016 when voice assistants like Alexa and Google Assistant were breaking out into mainstream use. By May 2018, there were 300,000 different chatbots available on Facebook Messenger alone, most of which were bots made for handling basic customer service tasks.
This time, however, the chatbot experience has gotten a significant upgrade thanks to the recent advances in generative AI. Meta is certainly aware that this Meta AI can do so much more than providing pre-determined answers based on pre-set trigger words. The previous generation of chatbots was purely functional, but this new wave of generative AI-powered chatbots is here to create and entertain.
With this Meta AI, Meta is set to leverage the scale of its popular apps to make generative AI more accessible than ever. It would reach hundreds of millions of people who will likely be surprised by how good the answers are, especially compared to the previous generation of chatbots. Plus, the Microsoft partnership will continue to shore up Bing in the ongoing AI search arms race. If the Meta AI takes off and becomes an emerging search channel, Meta’s massive user base could mean further damage to Google’s dominance in the search market.
AI Personas: Everyone, Everywhere, All At Once
Of course, Facebook is in the engagement business, and just one general AI wouldn’t be much fun. To that end, Meta smartly recognizes that there is nothing more engaging than media personalities, which is the bedrock of the modern media business that drives attention and engagement. From reality TV contestants to TikTok creators, having a strong, memorable personality is a prerequisite to standing out in today’s fragmented media landscape and attracting attention (and ad dollars).
Beyond the general-use Meta AI, Meta has also created specific-function AI chatbots out of its LlaMA model. Designated by their domain of expertise, users can chat with AI bots such as Max the sous chef, Lily the copyeditor, or Lorena the travel agent to, respectively, ask for recipe recommendations, check spelling and grammar, or ideate a travel itinerary. Unlike Meta AI, each bot is built on the likeness of a human actor, and the existence of a 3D animated persona adds a lot to the perceived personality of each bot.
Then there are the celebrity AI personas, or “AI characters” as Meta refers to them. Working with a long list of household names like Tom Brady, Snoop Dogg, Kendall Jenner, and Paris Hilton, along with internet-famous creators like Mr. Beast and Charli D’Amelio, Meta created different characters for these celebrities to embody, each represented with a 3D animated version of the celebrities playing the said characters. All these AI characters can be summoned into conversations with users across all of Meta’s various platforms, where they will give you fun and helpful advice in their areas of expertise. Some of their assigned expertise is close to what the celebrities do in real life (Tom Brady’s character Bru can chat about sports) while others are seemingly random, as is the case with Snoop Dogg playing a Dungeons & Dragons dungeon master.
Regardless, what they say is less important than who is saying it. Thanks to Meta AI, their responses are delivered via text alongside some facial expressions and hand gestures from 3D animated characters based on the likeness of said celebrity. Similar to the way that people would pay to go to Madame Tussauds and take selfies with celebrity wax figures, Meta is hoping that the AI-generated synthetic version of celebrities will draw in the fans and boost engagement across its platforms, whether its chat apps or in VR environments (more on that later).
The logical end of such deployment of generative AI is, of course, to allow everyone to create an AI person based on themselves. Meta announced it is working on an AI Studio feature in Alpha testing, which will not only let users generate 3D avatars in Horizon Worlds, but also allow businesses to create AI personas that “reflect their brand’s values and improve customer service experiences.” One day soon, you might be talking to an AI version of Jake from State Farm or Flo from Progressive on Facebook Messenger, having a lovely chit-chat before asking them for a quote for your home insurance.
For creators, the ability to create an AI version of yourself that is ready to engage and chat 24/7 spells some fascinating implications. On one hand, the parasocial relationship that a lot of popular content creators already cultivate with their followers can be taken to a whole new level, with followers having access to a synthetic version of their favorite online personalities all the time. On the other hand, doing so would also raise the age-old question of authenticity in digital-native connections. Would Meta’s AI model be good enough to escape the uncanny valley and fully engage with fans on behalf of the creators, or would it be janky and inauthentic to degrade the connection between creators and their fans? As Casey Newton of the Platformer newsletter points out, we might be witnessing the beginning of a partially synthetic social network here.
In addition, what would Meta’s business model be in order to scale such a feature to all creators on its platforms? Would it be a subscription model, or would it be subsidized by ads? After all, the cost of running the servers for LLMs is not cheap. So far, all the major tech companies are moving fast to roll out consumer-facing AI products without much thought on viable business models that could sustain their operations. And it doesn’t look like Meta has thought that far ahead yet.
How AI Boosts Meta’s Metaverse Dream
As Mark Zuckerberg noted in the keynote, this Connect event was about the intersection of AI advances and the next computing platform. Despite spending years to get Oculus VR headsets off the ground, VR remains a niche console business, primarily aimed at immersive gaming. The Ray-Ban Stories, Meta’s first generation of camera-equipped sunglasses made in collaboration with the sunglasses brand, also sank like a stone following its release in 2021. It would seem like Meta is having no luck with the mixed reality devices it has spent tons of R&D resources on over the years. Yet, the recent advances in generative AI may just point to a new UI design for Meta’s head-mounted wearables.
The new Oculus Quest 3 was one of the first products announced at Wednesday’s keynote, and it very much felt like Meta simply wanted to get it out of the way first to make room for the more exciting AI announcements. It is interesting to note that Meta’s demos placed special emphasis on the headset’s improved passthrough tech, which enables it to deliver an experience similar to how Apple’s Vision Pro works — a technically VR headset simulating an AR-like user experience. Of course, it would be unrealistic to expect a $500 headset to deliver a user experience on par with the $3,500 Apple device. But if the Oculus Quest 3 can get even close to what Apple is promising with Vision Pro, then next year’s VR market might look very interesting.
The key way that the new Oculus headset integrates with Meta’s new AI experiences is through the aforementioned AI characters, allowing users to interact with celebrity AIs in immersive environments. Users can also chat with Meta AI for information or simply companionship. Overall, nothing too drastically different from the kind of existing VR experiences that Oculus headsets already deliver.
In contrast, the other Meta headset product had a much deeper integration with AI. Besides the expected video recording and live streaming features, the new Ray-Ban Meta smart glasses (no more “Stories”; still a mouthful) will ship with Meta AI integrated on device, which will allow users to interact with it completely hands-free for a unique “on-the-go” wearable experience. Users can simply talk to the built-in Meta AI on the smart glasses by saying “Hey Meta”, and while it can only offer text-based responses at the moment, voice response will be coming soon next year along with visual recognition, which should unlock interesting new use cases for it. Priced at a relatively low $299, Meta is clearly placing its bet on AI-powered wearables to be the next computing platform, and it wants to be the one leading the charge this time.
Of course, all this comes back to Mark Zuckerberg’s undying dream of building a metaverse platform. But instead of betting solely on Horizon Worlds on Oculus headsets, the recent AI gold rush has seemingly made the Meta leader realize that there are other paths to the metaverse. Earlier this month, Meta finally expanded access to Horizon Worlds to mobile and web browsers. Now with the launch of this pair of smart glasses that puts its AI assistant front and center (right in your face, one may say) in its user experience, Meta is trying to make the metaverse part of our existing reality by familiarizing users with AI interactions and AI personas, while also steadily rolling out AI-powered creative tools to users to allow everyone to build a part of the metaverse.
For brands, this remains a wait-and-see moment. As I previously mentioned, all the major tech companies are rushing to come out with consumer-facing AI products without much thought on viable business models to pay for the high cost of running LLMs at scale. And without a clear business model, it would be difficult to discern where the brand opportunities lie. Meta did tease about an enterprise version of the Oculus Quest 3 made for productivity use cases, but that is far less important than figuring out how your brand messages would show up via Meta AI, and what kind of branded AI personas your brand might need to generate down the road. For now, get your 3D branded asset ready, and start experimenting with how this new generation of AI chatbots can help you engage with customers on a more personal level.
