Recently, some Facebook users have reported receiving strange automated messages from Facebook containing phrases like “Are you having a good day?” and “Hope you’re doing well!” Understandably, many users find these unprompted messages odd or even creepy. So why is Facebook sending these automated messages in the first place?
There are a few potential reasons why Facebook may be experimenting with automated messaging:
To test AI conversational agents
Facebook has invested heavily in developing artificial intelligence (AI) for conversational agents (chatbots) in recent years. They likely want to test out their chatbot capabilities at scale by deploying them on real Facebook users. The innocuous “hope you’re doing well” messages could be Facebook’s way of testing basic chatbot functionalities on their platform.
To improve user engagement
Facebook profits primarily from ads, which depend on users actively engaging with the platform. Messaging users directly could prompt them to open the app and interact more, boosting Facebook’s key engagement metrics. If users respond to the chatbot messages, it may signal to Facebook that they enjoy conversational features and lead to more personalization down the line.
To gather data for training algorithms
Any responses that users send back to the chatbots provides valuable data for Facebook to improve its AI systems. Facebook has access to vast troves of user data, but actual conversational data is lacking. The automated messages could be an attempt to gather real human conversational data at scale to train algorithms on. Any responses teach Facebook’s systems how real users talk casually on the platform.
Examples of the Automated Messaging
Here are some examples of the types of automated messages Facebook appears to be sending some users in recent weeks:
– “Hi! Are you having a good day so far?”
– “Hello! Hope your day is going well!”
– “Hey there! Sending you some positive vibes today!”
The messages are conversational but remain non-specific and innocuous. They do not reference any user details or activity. The vagueness likely indicates that these are initial tests rather than personalized messages.
Some users have received the messages through Facebook Messenger while scrolling through their News Feeds. Others have gotten them as notifications from the sidebar chat box feature. This suggests Facebook is testing various surfaces for deploying their conversational agents.
Message Frequency
In most cases, users receive just a single automated message from Facebook. However, some have reported receiving multiple messages over the span of a few days or weeks.
The frequency range seems to indicate that Facebook is still experimenting with optimal timing and quantity. Too many messages may annoy users, while too few will not provide enough data.
Geographic Targeting
Thus far, reports of these automated Facebook messages primarily come from users in the United States. However, some users in the U.K. and Canada also claim to have received them.
This limited geographic distribution likely means that Facebook is still only trialing the automated messaging in a few test countries. A wider international rollout may follow if these initial tests prove successful.
User Reactions
User reactions to receiving the automated Facebook messages seem mixed:
– Some find them fun and will happily engage the chatbot in brief conversation.
– Others are neutral to the messages, neither responding nor complaining.
– Many users consider the messages odd and invasive, even if they know it’s an AI system.
Here are some example responses that users have posted from their interactions with the chatbots:
| User Reaction | Sample Message |
|---|---|
| Positive | “Haha it was a pretty good day, thanks for checking Facebook!” |
| Neutral | No response |
| Negative | “This seems unnecessary, please don’t just message me out of the blue.” |
While some take the messages in stride, others clearly consider the unprompted messages invasive or creepy. Facebook will need to tread carefully as they expand automated messaging to avoid annoying users.
Privacy Concerns
A major source of user frustration seems to stem from privacy concerns. Facebook has faced backlash over privacy issues many times before. These unprompted chatbot messages strike some users as overly invasive and tone-deaf given Facebook’s past PR problems.
Many users complain that the messages did not come with any notice about their purpose or ability to opt-out. Greater transparency may help ease user concerns about the automated messaging.
An “opt in” measure could also give users more control. For example, a pop-up that explains the chatbot pilot program and allows users to consent or decline to participate.
The Future of Facebook’s Automated Messaging
It remains unclear whether Facebook plans to expand the AI chatbot messaging further. Here are some possibilities:
Personalized Messages
If the broad test messages show positive results, Facebook may try more personalized messaging. For example, referencing users’ posts, friends, pages followed, or recent activity. This could make the experience less generic and more “human.”
New Features and Capabilities
Facebook may enable new features like users setting their chatbot avatar, persona, voice, or other customizations. Chatbots may also integrate with other Facebook services like event recommendations or friend suggestions.
Commercial Use Cases
Facebook could offer conversational messaging capabilities to businesses and advertisers. For example, a clothing brand bot could message users about new sales and product launches. However, this may spark even greater backlash without proper opt-in measures.
Expansion to WhatsApp and Instagram
Facebook also owns WhatsApp and Instagram, so they may deploy automated messaging on those platforms too. WhatsApp in particular emphasizes conversational UI and has over 2 billion users.
Recommendations for Facebook
If Facebook does decide to make automated messaging a bigger part of their platforms, here are some recommendations:
Increase Transparency
Explain why users are receiving the messages, whether it’s an AI bot, and privacy/opt-out policies. This will reduce confusion and mistrust.
Make Participation Optional
An opt-in program would be far less invasive than forcing messages on users without consent. This places control back in users’ hands.
Gather User Feedback
Actively survey users on whether they enjoy the automated messages and find them useful. Feedback can inform improvements and new capabilities.
Refine the AI Capabilities
Work on making conversations feel more natural and personalized to each user’s interests and relationships on the platform.
Proceed Cautiously on Commercial Use
Tread very carefully with regards to enabling advertisers and businesses to message users without explicit opt-in consent, as intrusive messaging will exacerbate mistrust.
Conclusion
In summary, Facebook appears to be trialing automated conversational messaging powered by AI chatbots. The broad tests seem aimed at gauging user reception and gathering data to improve the technology. It remains to be seen whether Facebook will expand the messaging further. User reactions are currently mixed, highlighting the need for transparency and optional participation if Facebook decides to make chatbots a bigger part of their messaging strategy. Careful, consensual implementation will be key to creating a useful and non-invasive experience. Facebook faces a tricky balancing act between data collection imperatives and user privacy concerns.