Generative AI video: Excitement tempered with scepticism

Google unveiled a new generative video AI model—Veo which a direct salvo against the younger upstart and rival, OpenAI’s Sora. (AP Photo/Jeff Chiu) (AP)
Google unveiled a new generative video AI model—Veo which a direct salvo against the younger upstart and rival, OpenAI’s Sora. (AP Photo/Jeff Chiu) (AP)

Summary

  • Generative AI video models Google Veo and OpenAI Sora can seem quite magical, but their adoption may trigger copyright concerns and debates

New Delhi: A week ago, Google unveiled Veo, a new generative AI video model. In some ways, this was the search giant's direct salvo at younger upstart and rival, OpenAI’s Sora, which was introduced this February. 

Both claim to generate cinematic-grade video clips with just a few words of text. While this has sparked debate around AI’s impact on creative industries such as filmmaking and advertising, early reactions to their much-touted ‘magical’ abilities are lukewarm.

According to three content creators and three filmmaking and advertising veterans that Mint spoke with, although neither Veo nor Sora are available to try in India currently to test properly and raise a verdict, they appear a bit underwhelming, and may not become mainstays for video production.

Instead, the footage generated by such AI models may at best be used to visualize scripts at the pre-production stage, or create filler or ‘B-roll’ content that is used for background purposes in longer, more expansive movies and advertisements.

Other GenAI Video AI models

To be sure, Sora and Veo are not the only ones. On 27 January, Google Research unveiled Lumiere—a new text-based AI model that claimed to “portray realistic, diverse and coherent motion". 

Beyond Big Tech, startups have made generative video tools available to the general public to fiddle with for more than a year now—from UK-based Synthesia’s Express-1 model to New York-headquartered Runway’s Gen-2 model. Google, meanwhile, invested in Runway’s $141-million funding round last June. These show serious potential for generative AI video. 

Both OpenAI and Google claim that multi-modality of their video generation models—the ability to support input prompts in more than one format, including text, audio or images—is a key ability. Given Big Tech’s might and vast resources of computing and data available at hand, this can make for a serious resource for creative professionals to either leverage, or contend with. 

Also read Week in tech: Generative AI is coming to your Google search

For instance, Bengaluru-based technology content creator Varun Mayya, who has a cumulative following of nearly 2.2 million viewers across his multiple Instagram and YouTube channels, uses AI to generate filler footage as well as piece-to-camera videos for many of his content. 

While Mayya focuses on AI-based educational video content for developers and engineers, he has also used AI to gain assistance in his video production schedule.

Unlike Mayya, many others keep away. Two other content creators, who work closely with YouTube and requested anonymity to protect their respective brand partnerships, expressed that generative video platforms may not be comprehensive and could lead to branding complications. 

“Brands may not prefer knowing that any specific video was used with generative AI—there is a fear in the market that using AI videos could lead to copyright lawsuits, and no commercial entity wants such complications. This is a hindrance right now in India, at least until there is regulatory and legal clarity," one of the creators told Mint.

Each of these creators, however, confirmed that generative AI is well and truly being used in India at more than an experimental pace. While the early mood is anticipatory, it is also lukewarm—some cite quality issues, while others are concerned about intellectual property infringements.

“Most content creators sign originality agreements with clients. For them, if the material used to train a generative AI model belongs to a third party, and this reflects in the output, this could lead to an infringement of intellectual property," said Shailendra Bhandare, partner at law firm Khaitan & Co.

Bhandare added that multiple clients may get the same results if they use the same third-party material, which could lead to breach of contract and infringement of third-party IP.

Content oversaturation

Creative veterans, too, have a mixed view. Yash Kulshresth, chief creative officer at Mumbai-based creative agency Atom Network, said that most AI tools so far, based on initial inspection and review, “lack the emotional depth and unique perspective that humans bring".

“On the positive side, it can make visualizing storyboards and concepts more efficient and cost-effective. This democratization of creativity can enable smaller agencies and independent creators to compete more effectively and foster a new era of rapid prototyping and iterative feedback. It can also serve as a good alternative to regular hygiene content pieces for brands to put up on social media," Kulshresth said.

However, he added that content oversaturation is a significant risk. “The ease of generating visuals might lead to an abundance of mediocre content. This could make it more challenging for truly innovative ideas to stand out. The key will be for creatives to leverage AI as a tool that enhances their work, focusing on originality, authenticity, and emotional connection to maintain the distinctiveness of human-driven creativity," he said.

Death of creativity?

Despite this, all parties acknowledge that AI is an evolving field, which makes it difficult to predict its long-term impact. On 22 April, Synthesia unveiled its latest AI model—Express-1. The latter showed the ability to generate avatars that could mimic human speech, thus potentially replacing an anchor narrating a story on screen. Synthesia is also quite serious about its work—it is part of the Coalition for Content Provenance and Authenticity (C2PA) body that includes Google, Microsoft and OpenAI.

Fellow upstart Runway unveiled its latest model, Gen-2, in March last year. Since then, the company has showcased various video-generating capabilities of its AI model, and has even established an ‘AI Film Festival’ in Los Angeles with Nvidia as a partner. On 15 April, the company announced 10 winners of films made using its AI model and tool—offering nearly $35,000 in cash prizes.

Also read OpenAI releases GPT-4o to power ChatGPT, Altman says feels like AI from movies

This clearly reflects AI’s potential. “We have never seen something like this. It would be naive to think of our current response as cast in stone. It's just the beginning of AI and, ultimately, the future of our industry will be defined by the synergy between human creativity and technological advancements, ensuring that the heart and soul of our work remain unmistakably human," Kulshresth added.

The one thing most creative professionals believe, however, is that AI will not be the death of creativity. 

“A fundamental thing to understand is that AI does not have its own imagination," said Chaitanya Chinchlikar, vice-president and chief technology officer at Mumbai-based Whistling Woods International Film School. 

"At the level of every pixel that AI creates, it will be a synthesis of content that an algorithm has already seen, or is trained upon. This will make AI an assisting tool at best for the visual creation industry, from which some segments of the industry might benefit. But this will certainly not take away jobs from creative professionals."

Also Read: Can generative AI create a new religion that all could embrace?

 

Catch all the Business News, Market News, Breaking News Events and Latest News Updates on Live Mint. Download The Mint News App to get Daily Market Updates.
more

topics

MINT SPECIALS