NExT-GPT: Revolutionizing AI with Multi-Format Outputs, Text, Image, Audio & Video
The world of artificial intelligence (AI) is experiencing a revolutionary shift with the introduction of NExT-GPT, a large language model (LLM) developed by researchers from the National University of Singapore and Tsinghua University. Unlike its predecessors, NExT-GPT goes beyond pure text output and offers a wide range of capabilities including text, image, audio, and video outputs.
While giants like ChatGPT and Google Bard dominate the AI scene, NExT-GPT aims to disrupt the status quo. Known as an any-to-any system, it can process inputs in various formats and deliver responses in the desired output format, be it video, audio, image, or text. In practical terms, this means that users can input a text prompt and receive a video response, or input an image and get an audio output.
Although ChatGPT recently announced its ability to see, hear, and speak, NExT-GPT takes this concept to the next level by offering video capabilities as well. While several alternatives and rivals to ChatGPT have emerged in the past year, very few LLMs have been able to match its text-based output while offering additional features. Interested individuals can try out NExT-GPT on its GitHub page or demo site.
Having experimented with NExT-GPT on the demo site, I am impressed, though not completely blown away. It is worth noting that this is still a work in progress and lacks the advantages of public feedback and multiple updates. Nevertheless, the results are quite good.
For instance, I asked NExT-GPT to transform a photo of my cat, Miso, into an image of him as a librarian. Although it may not reach the same level of quality as established image generators like Midjourney or Stable Diffusion, the resulting image was undeniably cute and satisfactory.
On the other hand, my experience with the video and audio features was not as successful. The generated videos had the typical made by AI look, with some distortion and irregularities. While the results were not terrible, they did fall short of achieving a seamless and realistic output.
Overall, NExT-GPT shows immense potential in filling the gaps in audio and video capabilities within established AI models such as those developed by OpenAI and Google. With further advancements, it is hoped that NExT-GPT will produce higher-quality outputs, enabling users to effortlessly create home movies featuring their beloved pets or incorporating other creative elements.
