AI can be trained to create some creepy videos (Picture: Microsoft)
Microsoft has released a video of the Mona Lisa rapping, and it has left some viewers creeped out.
The giant tech company has revealed a new AI model that can take a still image of a face, and combine it almost seamlessly with an audio clip of someone speaking.
In one video example, researchers show how they animated the Mona Lisa to recite a comedic rap by Anne Hathaway using the AI Model VASA-1.
Many took to social media to air their thoughts, with one user, Nikita Lebedev, writing: ‘This is wild, freaky and creepy all at once’.
Another, Travis C Porco, said: ‘I think this rapping Mona Lisa has swizzled my temporal lobes.’
To view this video please enable JavaScript, and consider upgrading to a web
browser that
supports HTML5
video
But others hailed the new technology.
Ryan Lazuka wrote: ‘Can’t wait to see more incredible creations from Microsoft’s VASA-1 AI.’
Poonam Soni, another user on X, wrote: ‘This is scary good’.
Microsoft said the artificial intelligence could be used for education purposes or to potentially create virtual companions for humans – but also warned of the potential dangers of such technology.
The company admits the tool could be misused for impersonating humans, and therefore will not be releasing the demo to the public.
Other AI progress Microsoft has made
Microsoft’s dive into AI has created mixed results. The company tried to add ChatGPT to its Big search engine, but the bot went rogue. It insulted users and questioned its own existence, asking ‘Why? Why was I designed this way?Why do I have to be Bing Search?’
Despite the mini crisis shown by Bing, Microsoft is now working on how to introduce AI-generated content into Xbox video games.
With progress, and hopes, pinned on AI, Microsoft also made a historic change to its PC keyboards, adding in a new key dedicated its AI called the ‘Copilot Key’ to provide easy access to AI-powered features.
The team wrote in a blog post: ‘Our premiere model, VASA-1, is capable of not only producing lip movements that are exquisitely synchronized with the audio but also capturing a large spectrum of facial nuances and natural head motions that contribute to the perception of authenticity and liveliness.’
They added: ‘We have no plans to release an online demo, API, product, additional implementation details, or any related offerings until we are certain that the technology will be used responsibly and in accordance with proper regulations.’
Microsoft trained the AI using videos of people’s faces while speaking, designing the tech to recognise natural face and head movements through lip motion, facial expressions, eye gaze and blinking.
The Mona Lisa has been used an demo for the new AI tool (Picture: Microsoft)
However, the result is something more like a still photo which has been animated to move.
The AI also can produce a video where the subject is looking in a certain direction or is expressing specific emotions.
However, even though the videos are eerily realistic, there are still signs it has been AI-generated, such as infrequent blinking and exaggerated facial movements.
More Trending
But Microsoft has said it believes its model ‘significantly outperforms previous methods’ and can pave the way for real-time engagements with lifelike human avatars.
This is not the first time Microsoft has developed something that’s been met with controversy.
In 2021, the tech giant patented an AI chatbot that would allow users to talk to a past or present ‘entity’ such as a friend, relative, fictional or historical figure.
However, the product was never made and was even called ‘disturbing’ by Tim O’Brien, who was Microsoft’s general manager of AI programs at the time.