Is Open AI an existential threat to Google?

Plus the birth of text-to-3D...and other stories in Generative AI.

Dec 23, 2022
2

Hello and happy holidays to all,

My conversation about ‘the week in Generative AI ‘with my friend Henry Ajder was so popular, that we’re back for a bumper Christmas edition.

This week’s tidbits:

Riffusion: Real-time music generation

  • Real-time text-to-music generation is here with Riffusion.

  • Built on Stable Diffusion, it generated audio based on ‘Spectograms’ — or visual representations of sound.

AI-Composer Bias

Deep Voodoo secures $20m investment from CAA

  • The creators of Southpark have raised a $20m investment for their ‘deepfake VFX studio’ from CAA’s Connect Ventures.

  • Deep Voodoo created the deepfakes in the video for Kendrick Lamar’s critically-acclaimed hit, ‘The Heart Part 5.’

  • Intriguing to see the world’s biggest talent agency making a play in this space. An indicator of how AI will transform Hollywood.

(PS — I will be at CES in January to discuss ‘AI goes to Hollywood’ with SAG-AFTRA, the Hollywood actor’s guild. Give me a shout if you’re going to be in town!)

Open AI’s Point-E: Text-to-3D

  • The generative models keep coming….and Open AI keeps delivering.

  • This week, OpenAI released Point-E: a new text-to-3D model.

  • The model is open-source and has vast implications. 2023 will be a year for 3D.


Is Open AI an existential threat to Google?

AI-generated faces proliferate in ‘synthetic fraud.’

  • Fake social media accounts are sporting AI-generated profile pictures.

  • Meta says more than two-thirds of the influence operations it took down this year used AI-generated images.


Wishing everyone a relaxing Christmas break, and if your Christmas has been cancelled because you’re working on Generative AI: get in touch!

Namaste,

Nina