Universitas Airlangga Official Website

OpenAI introduces Sora, text-to-video tool, here’s what a UNAIR AI analyst says

OpenAI Sora
OpenAI Sora

UNAIR NEWS – OpenAI has released its latest Artificial Intelligence model, Sora, which can generate video from text prompts. Responding to the news, Data Science lecturer and AI analyst at UNAIR, Dr. Aziz Fajar S.Kom., M.Kom., revealed the significant potential of this technology in an exclusive interview with UNAIR NEWS on Monday, Feb 19, 2024.

“Sora has an enormous potential. It can be used for various purposes, such as generating promotional videos, educational videos, and more, much faster,” he said.

OpenAI’s Sora is the third AI model following DALL-E and GPT. Similar to ChatGPT, Aziz explained that Sora operates using a sequence of images. “Sora creates an image from the text prompt, and subsequent frames are generated based on the previously generated image, resulting in a video,” he said.

“If GPT learns grammar from text, Sora might learn how objects move, or physics, in the real world,” he added.

Some time ago, AI-generated a video of Will Smith eating spaghetti, which was widely trolled by netizens. In less than a year, AI has advanced to the point of creating ‘realistic’ videos like those produced by Sora.

Responding to this, Aziz stated that the rapid development of artificial intelligence is a result of the sophistication of the underlying technology.

“The most difficult challenge in the development of computer science is the specification of the devices used. Once the appropriate device technology is identified, AI progress can accelerate quickly,” he said.

Sora proves the rapid growth of knowledge and technology. Apart from providing positive impacts such as creating promotional and educational videos, OpenAI’s Sora also brings some negative effects that need to be anticipated. Combined with an AI voice generator, the potential for criminal activities may increase. However, Aziz advised not to worry because tools to detect AI-generated videos will be invented soon.

“Just as security systems in computer networks work based on existing generators, the AI development for detection is likely to rely on the available generators,” he explained.

The Data Science lecturer also believes that AI Sora is an alternative for video creation. There are still limitations such as the maximum duration of 60 seconds for the generated videos.

Author: Muhammad Naqsya Riwansia

Editor: Feri Fenoria

Read also

Data Science Technology students create Green Data Center concept

Get to know Universitas Airlangga’s FTMM Data Science Technology