Last month, OpenAI unveiled the generative video AI model, Sora. Showcasing its capabilities to create realistic videos based on short text prompts, OpenAI provided clips such as a woman walking through Tokyo’s neon-lit streets and a dog jumping between ledges.
Sora’s showcased abilities and potential have swiftly garnered significant attention across the tech, business, and investment sectors. OpenAI has framed the generative video capabilities as a progression in their overall roadmap. ChatGPT demonstrated breakthroughs in accurately understanding and generating human-like text, and Sora takes this a step further by creating a more complex content type, video, based on natural language understanding.
The logical progression of AI development and technological advancements will see Sora, according to OpenAI, being able to create increasingly refined, realistic creations with widening applications across different domains. The ramifications will be significant when combined with other technologies.
Potential Applications for Generative AI Are Plentiful
Sora is not the first example of generative video technology. In June last year, Google-backed AI startup Runway released Gen-2, a model that can generate videos from text prompts or existing images. In January this year, Google AI introduced a generative video diffusion model, Lumiere, which claims to produce “full-length, realistic, and coherent” videos in one go.
As reported by Nature, there is a range of potential benefits for generative AI tools like Sora. Tracy Harwood, a digital culture expert at De Montfort University in the UK, suggests that the technology could be a more accessible way of presenting dense text, such as academic papers. One of its most important applications could be visualising complex concepts to communicate research findings to laypeople.
Another potential application of the technology could be in healthcare, where generative video AI could help replace human clinicians for patient consultations. Some may find this unnerving, says Claire Malone, a science writer and consultant in the UK. However, others may find it convenient if they need to ask healthcare professionals specialist questions repeatedly throughout the day.
Generative video AI tools like Sora could help researchers to explore large datasets, says Malone. While generative AI can sift through code and perform basic research tasks, it can also “do more complex things like giving it data and asking it to make predictions.”
The Introduction of AI into the Realm of Film Production
The impact of Sora on industry practices and associated employment has been widely discussed on the analysis site, The Eurasian Review. Just as ChatGPT has disrupted industries, replacing jobs and altering the way many professions operate, the introduction of Sora could do the same. The concern has been most notably raised among professionals in the film industry.
US-based actor Tom Hanks has spoken about using AI technology to extend his career beyond his years. “If you are an ambitious young actor trying to figure out how to shape your career and you are told, ‘Sorry, but Tom Hanks is going to play all the parts forever.’ Does that give you pause in terms of pursuing acting as a career?” asks Dominic Lees, a researcher in creative AI and filmmaking at the University of Reading in the UK.
Last year, Hollywood screenwriters went on an industry-wide strike in protest against the rampant use of AI in screenwriting. The introduction of Sora could bring AI to the forefront of filmmaking and significantly impact filmmaking globally.
Of course, there will be new jobs created with advances in AI applications. However, if the knowledge and skill accumulation that makes humans unique is replaced by AI, the world will become a considerably different place. ChatGPT can pass the bar exam without attending law school, and Sora can make films without attending film school.
Navigating the Ethical, Legal, and Societal Implications
A key challenge for generative video technology is misinformation, says Harwood. “We could very quickly be flooded with a huge amount of incredibly convincing misinformation, and that’s really worrying.”
As reported by Scientific American, AI has the potential to copy or closely mimic copyrighted works and present them as original generative content.
The challenges of the digital age will be exacerbated as AI technology advances, claims The Eurasian Review. Imagine a world where billions of netizens could easily produce and share videos or films made with tools like Sora.
While the internet has made accessing, creating, and distributing information a “universal power,” it has not resulted in true information plurality. The era of information abundance has been accompanied by information monopolies and “effective information deserts.” Those in control of the underlying technologies and rules of the internet still control what information reaches audiences. For instance, advanced AI models and applications could be controlled by a small number of corporations, giving these AI giants another form of power that is not easily confined by law. If this trajectory continues, AI giants could become “unstoppable monsters.”
Generative video AI tools democratise the opportunity to be a creator of media content, but they also present society with broad challenges, says Nature. There needs to be a fundamental shift in how we evaluate the content we consume to cope with this radical change in consumption.
