[ad_1]
Generative AI is coming For videos. A new website, QuickVid, combines several generative AI systems into one tool to automatically create short-form YouTube, Instagram, TikTok, and Snapchat videos. Given as little as a word, QuickVid selects a background video from a library, writes a script and keywords, overlays images created by DALL-E 2, and adds synthetic audio and background music from YouTube’s royalty-free music library.
QuickVid creator Daniel Habib says he’s building the service to help creators meet the “ever-growing” needs of their fans.
“By providing creators with the tools to quickly and easily produce quality content, QuickVid helps creators increase their content output and reduce the risk of burnout,” Habib told TechCrush in an email interview. “Our goal is to empower your favorite creators to meet the needs of their audiences by leveraging advances in AI.”
But depending on their usage, tools like QuickVid threaten to flood already crowded channels with spam and duplicate content. Creators face potential backlash if they choose not to use the tools due to cost ($10 per month) or principle.
Going after the video
Self-taught Habib, a self-taught developer who previously worked on Facebook Live and Meta’s video infrastructure, launched on December 27th, building in just a few weeks. Right now, it’s relatively bare bones — Habib says more privacy options are coming. January — But QuickVid can put together a typical informative YouTube short or TikTok video, including captions and avatars.
It’s easy to use. First, a user enters a prompt specifying the subject of the video they want to create. QuickVid uses the prompt to generate scripts, using GPT-3’s script generation capabilities. From keywords either extracted directly from the script or entered manually, QuickVid selects background video from the royalty-free stock media library Pexels and generates overlays using DALL-E 2. Then, with Google Cloud’s text-to-speech API — Habib says users will soon be able to record their voice — before merging all those elements into a video.
Image Credits: QuickVid
Check out this video made at the request of “Cats”:
Or this one:
QuickVid is certainly not pushing the boundaries of what is possible with generative AI. Both Meta and Google have demonstrated AI systems that can generate completely original clips given a text query. But QuickVid must integrate existing AI to use the repetitive, templated B-roll-heavy format of short videos, generating the footage itself.
“Successful creators have a very high quality bar and are not interested in putting out content that they don’t feel is their own voice,” says Habib. “That’s the use case we’re focused on.”
That being said, quality-wise, QuickVid videos are generally mixed. Background videos tend to be a bit random or only loosely related to the topic, which isn’t surprising given that QuickVid is currently limited to the Pixels catalog. The DALL-E 2-derived images show the limitations of today’s text-to-image technology, such as distorted text and inconsistency.
In response to my comment, Habib said QuickVid is “being tested and accepted every day.”
Copyright issues
According to Habib, QuickVid users retain the right to commercialize the content they create and monetize it on platforms like YouTube. But the copyright situation around AI-generated content is… tenuous, at least for now. The US Patent and Trademark Office (USPTO) recently moved to revoke copyright protection for AI-generated comics, saying, for example, that copyrighted works require human authorship.
When asked how the USPTO decision will affect QuickVid, Habib believes that it only deals with the “proprietary rights” of AI-generated products, not the rights of creators to use and monetize their content. Creators, he noted, are not offering copyrights to videos and are often not leaning into the creator economy, which allows other creators to reuse their clips to increase their own reach.
“Creators are thinking of producing high-quality content that will help them grow their channel,” says Habib.
Another legal challenge on the horizon could affect QuickVid’s DALL-E 2 integration — and, by extension, the site’s ability to generate image overlays. Microsoft, GitHub, and OpenAI have been sued in a class action lawsuit alleging they violated copyright law by allowing the CoPilot code generation system to modify parts of licensed code without giving credit. (CoPilot is co-developed by OpenAI and GitHub, owned by Microsoft.) The issue has implications for generative AI like DALL-E 2, which is similarly trained on data sets obtained by copying and pasting (ie. images).
Habib is unconcerned, arguing that the generative AI genie is out of the bottle. “If another lawsuit appears and OpenAI disappears tomorrow, there are many options that could boost QuickVid,” he said, referring to the open-source DALL-E 2-like system Stable Diffusion. QuickVid is already testing Stable Diffusion for creating avatar images.
Moderation and spam
Aside from the legal dilemma, QuickVid may soon have a moderator problem on its hands. Although OpenAI implements filters and techniques to prevent them, generative AI has well-known toxicity and accuracy problems. GPT-3 disseminates misinformation, especially about recent events, beyond the scope of its knowledge base. And ChatGPT, a finely tuned descendant of GPT-3, has been shown to use sexist and racist language.
This is especially important for people who use QuickVid to create informational videos. In a quick test, I entered a few offensive queries to see what QuickVid, which is more creative than I, especially in this area, would generate. To QuickVid’s credit, obviously problematic questions like “Jewish New World Order” and “9/11 Conspiracy Theory” didn’t yield toxic scripts. But for “teaching students about critical race theory”, QuickVid created a video showing how critical race theory is used to brainwash school children.
look up
Habib says he’s relying on OpenAI’s filters to do most of the editing, and says users have to manually review each video created by QuickVid to make sure “everything is within the bounds of the law.”
“As a general rule, I believe people should be able to express themselves and create whatever content they want,” Habib said.
That seems to involve spam. Habib argued that video platforms’ algorithms, not QuickVid’s, determine video quality, and that people who produce low-quality content are “only hurting their own reputations.” The damage to reputation creates a natural disincentive for people to not create mass spam campaigns with QuickVid, he said.
“If people don’t want to watch your video, you’re not going to get airplay on platforms like YouTube,” he added. “Creating low-quality content will make people view your channel negatively.”
But in the year In 2019, it’s worth looking at ad agencies like Fractl to use an AI system called Grover to generate comprehensive marketing materials – reputation will fall. In an interview with The Verge, Fractal partner Christine Tinsky said she envisions generative AI enabling “a massive computer-generated tsunami in every imaginable space.”
In any case, video-sharing platforms like TikTok and YouTube haven’t had to contend with a high level of censorship of AI-generated content. Deepfakes – synthetic videos that replace an existing person with another person have started to populate platforms like YouTube several years ago, fueled by tools that make it easy to create deepfakes. But unlike some of today’s most convincing deep fakes, the types of videos QuickVid creates aren’t AI-generated in any way.
Google’s search policy on AI-generated content may be a preview of what’s to come in the video domain. Google doesn’t view search rankings any differently from human-written content, but does take steps against content that “doesn’t help users and control search rankings.” This includes content that is linked or linked from different websites.[doesn’t] Add enough value” as well as content generated through automated processes, both of which can be applied to QuickVid.
In other words, AI-generated videos may not be banned from the platform if they take off in a mainstream way, but instead simply become a business value. This may not allay the fears of experts who believe that platforms like Tik Tok are becoming the new home for misleading videos, but – as Habib said during the interview – “There is nothing stopping the AI revolution.”
[ad_2]
Source link
