If you've ever wondered what would happen if ChatGPT could create videos, you're about to get the answer.
Sora is a new ai video generation tool recently announced by OpenAI, the maker of Microsoft-backed ChatGPT. Although it is currently only available to cybersecurity experts who are testing it, Sora is expected to be released publicly soon.
While ai video generators already exist and are in use, the Sora-created video samples released so far show a level of photorealism never seen before. Furthermore, OpenAI has become a leader in the field of ai and its entry into generative video could have big implications for society, as well as for those of us tasked with guiding our students through this new and brave world of text, images and computer generated content. and now, movies.
As an educator, I can see many ways this technology can be helpful to my students, in addition to the potential obstacles and concerns. Of course, it remains to be seen how all of this will play out, but here's what we know about Sora and the implications of her for the teachers so far.
What is Sora?
Sora is a new text-to-video conversion model from OpenAI that is currently being tested by security experts before being made available to the general public. Educators, in particular, should remember not to confuse it with Overdrive's K12 Digital Library Tool of the same name.
“We are teaching ai to understand and simulate the physical world in motion, with the goal of training models that help people solve problems that require interaction with the real world.” OpenAI said in its post introducing Sora. “Sora can generate videos up to one minute long while maintaining visual quality and compliance with user instructions.”
Sora works with a combination of the diffusion model used by OpenAI's Dalle-3 imager, as well as elements of the GPT-4 model. ai-video/” target=”_blank” data-url=”https://www.wired.com/story/openai-sora-generative-ai-video/”>Wired reports. The examples that OpenAI has shared so far are quite impressive.
In a video created from a one-paragraph message, a woman walks down a rain-soaked Tokyo street flooded with neon lights. In another, the immediate “historical footage of California during the gold rush” led to a drone-style shot of a realistic-looking Western city with horses and people walking the dusty streets.
“Sora is capable of generating complex scenes with multiple characters, specific types of movement, and precise subject and background details,” OpenAI notes. “The model understands not only what the user has requested in the message, but also how those things exist in the physical world.”
What are Sora's limitations?
However, Sora is far from perfect. Like all ai models, he can make mistakes or “hallucinations” that may seem more astonishing and strange given how well they do other things. “He may have difficulty accurately simulating the physics of a complex scene and he may not understand specific instances of cause and effect,” notes OpenAI. “For example, a person may take a bite of a cookie, but afterward, the cookie may not have the bite mark.”
In another example shared by OpenAI, a video of a man running on a treadmill walks in the opposite direction. Obviously this is a mistake, but it looks pretty good. The clip reminds me of the work of Maurits Cornelis Escher and is one of the most interesting Sora videos that OpenAI has shared so far.
How much will Sora cost?
OpenAI has not released any pricing details. But based on other ai tools it has released, it seems unlikely that full access will be free; however, that remains to be seen.
What are some implications for teachers?
Let's focus on the positive first: there are many potential teaching applications for Sora.
As a writing teacher, it might be fun to use this tool with students. I can see this as an attractive, if somewhat gimmicky, way to help fiction writing students visualize scenes. By requiring specific prompts, it can also help anyone learn to write with clear, precise language.
Teachers of other subjects may find fun uses for it too. For example, I would love to see how it models photosynthesis or the movement of planets. Does it provide an accurate visual physics lesson or can your students point out the flaws?
Now about worries. We all know that ai has a tendency to “freak out” and can amplify the bias of those who created it in unintentional and unpredictable ways. This is concerning with text- and image-based ai, but the stakes are even higher with video.
OpenAI appears concerned about some of these issues and says it is working to check for bias and will provide an identification label on videos created with the tool to combat misinformation. Additionally, applications requesting extreme violence, sexual content, hate images, images of celebrities, or the intellectual property of others will be rejected.
But given the challenges of existing ai technology, it is difficult to imagine any level of bias that does not escape these safeguards. Google, one of OpenAI's rivals, for example, recently announced that it would discontinue its ai portrait generation tool after reports of historical inaccuracies. ai-tool-historical-inaccuracies-portraits-1.7122704″ target=”_blank” data-url=”https://www.cbc.ca/news/business/google-ai-tool-historical-inaccuracies-portraits-1.7122704″>including allegedly depicting Nazis as people of color.
Even if OpenAI can get around these obvious cases of inaccuracies and biases, there are types of biases that might be inherently unavoidable. What if a student asks Sora to create a beautiful person walking down the street? What beauty standard will the model use?
As with other ai tools, rather than ignore these challenges and concerns, I would recommend instructors embrace them and discuss them all with their students. ai is a fascinating, fun, powerful and often scary new technology. Let's learn to navigate it with our students.
