The Download: AI’s mind-blowing skills

February 16, 2024

TLDR: OpenAI has developed a generative video model called Sora that can turn text descriptions into high-definition film clips up to a minute long. The company has no immediate plans to release it publicly but is conducting extensive safety testing. Google DeepMind has unveiled a new version of its AI model Gemini that can work with larger amounts of data, including video, text, and images. The model demonstrated its ability to identify specific moments in a transcript and detect humor. Both developments hold significant potential but also face challenges, such as safety considerations for AI models and the need for further research into phage therapy.

OpenAI has created an impressive generative video model called Sora that can transform short text descriptions into detailed, high-definition film clips. Although the company has no immediate plans to release the model publicly, it has shared sample videos with MIT Technology Review. The development is part of the growing trend of text-to-video generation, and OpenAI is conducting extensive safety testing before making it available.

Google DeepMind has introduced the next generation of its AI model Gemini, which can handle larger amounts of data, such as video, text, and images. In a demonstration, the model was given the transcript of the Apollo moon landing mission and a hand-drawn sketch of a boot. Gemini successfully identified the moment in the transcript represented by the drawing and recognized moments of humor. However, caution has been advised when interpreting AI demos, as there are often limitations and potential risks associated with such technologies.

Phage therapy, which utilizes viruses to combat bacterial infections, shows promise but also highlights the need for further research. A patient with a persistent blood infection was successfully treated with phages after antibiotics failed. However, ultimately, the therapy also failed to completely eradicate the infection. While the therapy represents a positive development, there is still much to learn about its effectiveness and potential limitations.

Latest from Blog