I’m still not sure if people will be interested in Meta’s AI bots, which try to simulate interactions with real people within the app.
The company tried this in 2016 with a Messenger bot, which nobody paid any attention to, and then tried again last year with celebrity-like AI-powered bots, which also didn’t catch anyone’s attention and ultimately died out.
But now the company is testing bots built by creators and influencers that can replicate those interactions, allowing them to spend more time creating and less time responding to pointless chatter with their fans.
The problem is, these people owe their status to their fans, and directing them to an AI chatbot feels like cheating them, even if it responds in the conversational style of their choice. And it seems like a step backwards, because no matter how realistic and convincing an AI chatbot becomes, it won’t be real. So you’re selling a fantasy in an app that is based on real human interaction and engagement.
At least in Meta’s eyes, authenticity seems to be lost in favor of mock responses, but I don’t think Meta users would see it that way.
It’s possible that more valuable uses of chat AI will emerge in the future that could deliver real value rather than just simulated engagement.
During the company’s second-quarter earnings call today, Meta CEO Mark Zuckerberg described future applications of AI chatbots built by individual businesses that leverage increasingly adaptive and intelligent AI tools to facilitate faster ordering and customized engagement.
According to Zuckerberg:
“Our goal is to make it easy for any business to ingest their content and catalog into our AI agents, helping them increase sales and reduce costs. Once this works at scale, we expect to dramatically increase business messaging revenue.”
So, at least conceptually, Meta is aiming to develop a “business AI” that can access an entire catalogue and information of products and provide personalized customer service within the app.
That’s essentially what Meta wanted to achieve with its Messenger bot, which, as mentioned above, nobody actually used.
So based on precedent, the metrics here aren’t all that great, but with smarter AI technology and more people getting comfortable working with AI tools, this could be a good bet and provide Meta with another opportunity to integrate AI into its business offerings.
Either way, this is a better use of AI than simulating human involvement, and as tools like ChatGPT become more widely adopted, it’s only natural that options like this will become increasingly in demand.
Of course, Meta is integrating AI into its ad targeting and creation tools, and there are obvious benefits to building a smarter system based on how users respond in the app.
Meta’s vision is that advertisers will eventually only need to choose an ad objective and set a budget, and the system will handle everything else, including creating images from a catalog, inputting copy, launching ad variations, and targeting.
This is evolving and we are already seeing results, and the benefits here are more tangible because this is more removed from the user’s perspective, but for the most part I don’t think people actually want bots to replace human interactions within an app.
The transactional interactions are different, which could be an area of opportunity, but there feels to be a clear distinction between these two use cases, and that’s reflected in user responses.
Overall, social media apps are just that: social, and Meta needs to heed the lessons of past chatbots to avoid making the same mistakes.
