This text belongs to ‘Artificial’, the newsletter on artificial intelligence that La Vanguardia sends to its readers every Friday. If you want to receive it in your mailbox, sign up here.
There is no week without surprise or shock in the world of artificial intelligence. After the seismic arrival of Sora, OpenAI’s photorealistic video creation model from a text description, other new features arrive that complement it. The border between reality and fiction is almost completely blurred and will soon be indistinguishable. The speed at which advances occur, outside of regulations, contrasts with the lack of control imposed on a technology with strong power to manipulate facts.
Sora’s impact is still huge. To the point that the actor and producer Tyler Perry has canceled an investment of 800 million dollars in some Hollywood studios because he considers that this technology could mean a turnaround for the film industry, in which he believes that many jobs will disappear. . But there are more initiatives that make the audiovisual sector feel in the middle of a storm. It is not surprising. Another company, LTX Studios, has presented a platform that develops entire films with a multitude of details.
The Institute for Intelligent Computing, of the Chinese group Alibaba, yesterday presented EMO, an AI for the generation of portraits and expressive videos based on audio. The spectacular model only needs a single reference image and the audio that you want to apply to it to generate videos with facial expressions and head movements of an overwhelming reality. To demonstrate how well it works, its authors have put the woman created by Sora who two weeks ago surprised us walking through the streets of Tokyo singing a song by Dua Lipa, Mona Lisa performing a song by Miley Cyrus or Audrey Hepburn from Roman Holidays responding to an interview. Everything from a photo.
The photographs have precisely once again caused a new problem for Google. The generation of images of people from its Gemini AI has had to stop being available to the public because the company has taken its fight against bias too far, creating unrealistic images, such as black African doctors treating white people, Popes blacks (one of them, a woman) and other kinds of nonsense inconsistent with reality and history. The issue opens up an interesting debate about falling short or overstepping AI guardrails. Another controversy, splashes Elon Musk squarely. In a question about who impacted society more negatively, the owner of X or Hitler, the AI ??couldn’t decide which of the two was worse.
This has been the week when AI took off at the Mobile World Congress in Barcelona, ??with numerous applications of artificial intelligence to all areas of mobility technologies. Among the most spectacular examples, the company Etisalat has shown at the congress its Ameca robot, a feminine-looking humanoid that reasons with ChatGPT responses and makes gestures very close to humans with its face, hands and arms. The device has been designed to show human feelings such as sadness, joy, anger or happiness. At the moment, Ameca has not been provided with mechanisms to move like that used by other robots. But can you imagine it if it had mobility?
