ChatGPT is not all you need. A State of the Art Review of large Generative AI models

  • 2023-01-11 15:48:36
  • Roberto Gozalo-Brizuela, Eduardo C. Garrido-Merchan
  • 64

Abstract

During the last two years there has been a plethora of large generativemodels such as ChatGPT or Stable Diffusion that have been published.Concretely, these models are able to perform tasks such as being a generalquestion and answering system or automatically creating artistic images thatare revolutionizing several sectors. Consequently, the implications that thesegenerative models have in the industry and society are enormous, as several jobpositions may be transformed. For example, Generative AI is capable oftransforming effectively and creatively texts to images, like the DALLE-2model; text to 3D images, like the Dreamfusion model; images to text, like theFlamingo model; texts to video, like the Phenaki model; texts to audio, likethe AudioLM model; texts to other texts, like ChatGPT; texts to code, like theCodex model; texts to scientific texts, like the Galactica model or even createalgorithms like AlphaTensor. This work consists on an attempt to describe in aconcise way the main models are sectors that are affected by generative AI andto provide a taxonomy of the main generative models published recently.

 

Quick Read (beta)

loading the full paper ...