Trish, I love your AI posts. I saw OpenAI just released Sora, can you talk about it?
Hi there, and Happy end-of-February (can you believe the month is almost over?!)! I hope you’re all doing well and having a great start to the week.
Thank you so much to this week’s question-er for the timely question (and the kind compliment! Love that you all are loving the AI content). You’re absolutely right – if you haven’t heard, OpenAI, arguably the world’s leading AI organization, recently released a new generative AI tool – Sora. Already, Sora is taking the AI world by storm and has left folks stunned, amazed, and yes, concerned (if you’ve been reading my other AI content, that should be familiar). But why? What’s the big hype around Sora? I’m more than happy to talk about it – and that’s exactly what I do in this post. Below, I offer y’all a brief introductory look at 1) what Sora is, 2) how it works, and 3) how it might change our world. Let’s get into it:
First up…what is Sora? Sora is OpenAI’s latest generative AI tool, a tool that, simply put, leverages AI to turn simple text prompts into beautiful videos and graphics. These scenes can be both realistic or imaginative – and they can be complex. You can specify characters, details, even the way that the camera is framing the shot, e.g., “The camera moves through the street.” There are some limitations, of course – for instance, videos can currently be up to a minute long. Sora is also not yet available to the public-at-large – only to folks who have been tasked with assessing it for potential harms and to some artists and designers, who are offering feedback – but no doubt, it represents the advent of a new age in film, video, and art. Whatever your thoughts are on AI, Sora is definitely impressive. (So hats off to OpenAI!) With that said, Sora is not the only text-to-video generation tool; Meta, Google, and plenty of other start-ups have their own tools in the works. So expect that Sora will be one of many tools competing for our prompts – our visual attention – in the coming years.
So, very briefly, how does Sora actually work? Put very simply, Sora (like all AI models) is a model that has been trained on tons and tons of data, in this case, video data. It is able to use what it’s learned from this data to generate videos based on the prompts you enter. You might be wondering…does Sora actually understand what it is producing? An early look at the model suggests that Sora doesn’t understand quite as well as you might think. Early reviews, for instance, note Sora generating videos where a person takes a bite out of a cookie…but then there’s no “bite mark” in the cookie. Similarly, Sora doesn’t necessarily understand the physical laws of our world – which, of course, limits how effectively it can mimic real life. But there’s good reason to believe that the technology will improve – OpenAI, and again, many other competitors, are hard at work doing just that!
Okay – now we have a good sense of what Sora is and how it works. But why should we care? How might Sora change our world? Well, it’s easy to imagine how disruptive a technology like Sora might be. For instance, you might wonder, is the age of human-made television and movies over? Will we ever have films with human actors again? You certainly wouldn’t be the only one thinking that…soon after Sora’s release, Tyler Perry, a notable American actor and filmmaker, put a planned $800M studio expansion on hold, noting that the technology likely meant that “jobs are going to be lost” in the future. So, at the very least, Sora will change what and in what ways video content is produced. There’s also plenty of concern about mis/disinformation. While OpenAI is currently working to minimize safety issues, it’s not unlikely that Sora will be used to generate inaccurate, fake – but extremely realistic – depictions, content that can be used to bully or harass folks or even to influence elections. Thus far, OpenAI has tried to tackle mis/disinformation by adding watermarks to the content its models generate, but even it itself has noted that these watermarks are very easy to remove (yikes).
I hope that that was an informative look at Sora! I’ll stay tuned (and keep y’all updated!) as this tool develops and continues to make waves.
Last thing…when you have a moment, do me a huge favor and share your internet-related thoughts and questions with me here. Remember, it takes 1 minute and absolutely anything you’re wondering about is valid. I look forward to hearing from you!
Thanks so much for tuning into this month’s last Ask Trish post. Have a great start to March,
Trish