Apparently people are posting less on Facebook and Instagram, so Meta is adding a new feature: an artificial intelligence (AI) bot, and you can chat with it.
Meta AI is being deployed in Facebook, Instagram, Messenger and WhatsApp. It’s essentially a conversation interface to which you can ask questions and request synthetic images, like OpenAI’s ChatGPT, Microsoft’s Co-Pilot and Google’s Gemini.
Artificial intelligence has been ubiquitous for the past year, but for many people, this will be their first interaction with the technology. Meta’s applications have billions of users: anyone who has managed to avoid AI so far will have it under their noses before long.
Why is this AI robot here?
The technology is recent and its usefulness is questionable. But big tech companies have decided that, like voice assistants and vertical-scrolling videos, AI is the next big thing. Facebook and Instagram use friends, family and groups to keep users’ attention. But these platforms are aging, and Meta may be hoping that a talking robot can replace some of the human interaction.
How to find it?
The Meta AI bot is integrated into the Meta apps search engine and messaging. It may also appear in your News Feed under certain messages. If you don’t yet see the AI icon – a thin blue ring – come back later, the deployment is gradual and happens more quickly on the phone app. Meta AI is also accessible on the web (meta.ai). It is not included in the Messenger Kids children’s app.
On Facebook, tap the search icon at the top: the search bar has been replaced with a bar that says Ask Meta AI anything. Anything with a blue circle brings up Meta AI’s chat interface. You can also tap the messages icon and chat with Meta AI as if it were a Facebook friend. Below a post in your News Feed, it will suggest questions to ask related to the post.
Meta AI also appears in search bars on Instagram, Messenger and WhatsApp. If your accounts are linked, the Meta AI conversation should pick up where you left off, no matter which app you enable.
How to deactivate it?
That’s impossible in research, Meta said. In WhatsApp there is an option that hides the new Meta AI button (Settings-Chats-Show Meta AI Button). But it stays in the search bar. Other apps have an option that mutes responses. When Meta AI is asked how to disable it, it gives incorrect answers, with instructions that don’t work and settings that don’t exist.
You can delete a chat with Meta AI from Recent Chats like any other chat: swipe left and select DELETE on Instagram, More-Delete on Facebook and Messenger, and More-Delete THE cat on Whatsapp.
How to start?
Type full sentences or random words into app search bars or directly into Meta AI. Start by asking simple questions and even ask for a list of ways to use it.
(At the time of publishing this article, Meta AI understood the questions in French and responded in English. When asked to repeat the answers in French, a response in French appeared, but immediately disappeared, replaced by a message indicating that the robot could not does not yet speak French, but that he is working on it and that a message will be sent when the French functionality is ready.)
What can it be used for?
An AI robot is like an enthusiastic but unreliable friend. You can ask him anything, but you should never trust the answer. With this in mind, use Meta AI for fun and non-critical tasks. Ask random questions like you would Google, start conversations to make you feel less alone, and use them to brainstorm ideas.
Meta AI can also generate images, although, according to our tests, they have the characteristic flaws of AI: hyperrealistic lighting, errors in the eyes and fingers and, often, women with generous and conspicuous cleavage .
IMAGE GENERATED BY META AI
Meta AI can generate images, but they have the characteristic flaws of AI.
But we can try many other things. Ask Meta AI to animate images, a summary of today’s news or ask it to write to you as if it were a specific character. Since Meta AI is integrated with other Meta products, you can ask it to “Show me clips of people learning to skate” (results vary).
To avoid bland answers, ask follow-up questions and provide plenty of detail.
In what context should you never use it?
Don’t use the AI as an authority for anything important (medical advice, schoolwork, work). Ethics forbids using it for school work, but Meta AI can produce essays full of commonplaces and clichés.
Experts warn that there is a risk of misinformation with AI tools. Better not to use it for anything sensitive or political. Instead, turn to human sources like journalists, experts, even Wikipedia and Reddit, before AI.
How is it different from other AI robots?
For the most part, Meta AI seems to give the same generic answers as its competitors. Asking five different conversationalists about the best taqueria in San Francisco, a vegetarian meal plan, the existence of God, and how to know if polyamory is a good idea for you yields incredibly similar, mundane, yet neutral answers. .
Does it keep my information?
Take the same precautions when writing to an AI bot as you do when searching on Google. Meta records everything, the data is anonymized to protect privacy; therefore, nothing is linked to your name or identity. This is the norm in all technology companies, but according to experts, it is possible to trace back to people using other data. If you want to delete a conversation, you can use the shortcut
“/reset-ai” and Meta says it will delete the conversation from its servers.
Read this article in its original version
(in English; subscription required)