The Rise of AI Content: A Cause for Concern
It may sound dramatic, but it’s crucial to adopt a mindset where you question the authenticity of everything encountered online.
While the internet still hosts a plethora of content created by actual people, the rapid advancement of AI-generated media blurs the line between reality and fabrication. This makes it a disadvantage to assume that the content filling your social media feeds is genuine.
Don’t dismiss this article just because you believe you can spot AI content. Although the current outputs from your social media algorithm might seem easy to identify if you’re vigilant, the next wave of AI-generated videos is here and far more sophisticated than you might expect.
AI Content’s Deceptive Nature
Many individuals recognize the hallmark characteristics of AI-generated videos. For instance, a viral clip depicting a cat parent rescuing its kitten from a burning plane stands out as unmistakably artificial to most onlookers. Similarly, the notion that former President Trump is operating a construction site is absurd, and it’s evident that videos depicting cat farming families are fabricated.
However, certain videos can easily deceive those less familiar with AI technologies. While some viewers may discern the artificiality in a clip of babies dancing, countless comments indicate that many others are unaware. Even seemingly benign clips, like pets reacting to a bird and a toy alligator, can leave viewers in doubt about their origin. The flood of AI-generated content, particularly in formats like America’s Got Talent, illustrates that remarkably realistic visuals can still sway large audiences.
While it’s concerning to witness how many of these seemingly plausible AI videos mislead viewers, a more profound worry looms on the horizon.
Currently, many AI videos rely primarily on visual appeal and background sounds to convey a sense of authenticity. Notably, characters in these videos often do not speak, and when they do, the result can feel disjointed due to misaligned lip movements and mechanical voices. This method allows AI creators to prioritize the lifelike depiction of people and animals, hoping that viewers will overlook something feels amiss.
OpenAI’s Sora video model showcased impressive realism last year, particularly with visuals of a woman “filming” her reflection in a train window. Yet, such videos typically lack fully rendered conversations, leading viewers to assume they are legitimate human-created media.
A Shift in AI Video Technology
A recent event has intensified concerns about the state of truthfulness on the internet. At the latest Google I/O conference, Google introduced Veo 3, its newest AI video model. Similar to competing technologies, Veo 3 can generate highly realistic sequences that were on display during the presentation. While the visuals were not groundbreaking, they certainly raised eyebrows.
What sets Veo 3 apart is its ability to generate accompanying audio, including sound effects and lip-synced dialogue. To showcase this feature, Google presented a clip of an old sailor at sea, complete with sharp video quality and fluidly synced speech. While some viewers may find nuances indicating its AI nature, many might easily be misled, especially fans of faux AGT content.
What genuinely sparked concern, however, was not just the initial demonstration but the content created by users after gaining access to Veo 3. Publications like PetaPixel have compiled impressive (and troubling) highlights of recent videos crafted with this technology.
For instance, a clip featuring a streamer playing Fortnite, with all elements generated by Google’s AI, exemplifies the impressive yet alarming potential of such technology.
In another instance, a video showcases concerts that never occurred, complete with synthetic musicians and nonexistent crowds. Every aspect, from vocals to instruments, was artificially generated.
Perhaps the most alarming video featured a fabricated auto show, complete with fake interviews of nonexistent spectators. Despite its imperfections, the superficial authenticity captured the attention of onlookers, showcasing how easily even the discerning might fall victim.
These visuals, dialogues, and settings present a veneer of realism that can easily deceive, particularly when it appears in the midst of scrolling through TikTok or Instagram. Even the previous Veo 2, which isn’t as advanced, offers realistic camera movements, providing creators the tools to produce content that feels lifelike. With the technology evolving continually, the quality of AI content is likely to improve.
A premium subscription for Google’s advanced AI video generation tools comes at a cost of $250 monthly, which may be significant but is attainable for many creators. An alternative plan at $20 monthly still grants access to Veo 2 and Flow, potentially ushering in a wave of lower-quality yet still deceiving outputs.
Embracing a Skeptical Perspective
No technology is without flaws, and it’s crucial to note that not every output from Veo 3 will look indistinguishable from real footage, nor will it be devoid of typical AI markers. Reports suggest peculiarities in Veo 3’s training datasets, revealing it produces the same odd “dad joke” upon request for stand-up performances.
This calls for a heightened alertness to discern the real from the artificial. When engaging with online content—particularly rapid-fire algorithmic videos—it’s wise to approach with suspicion, presuming, at least at first, that the material is fabricated unless proven otherwise. While this may seem extreme, the visual, aural, and even narrative nuances of the latest AI content leave little room for other approaches in an increasingly fabricated digital landscape.
The implications of this advancing technology are daunting. Today, it’s the fabricated performances of streamers and musicians. Tomorrow, it may be distorted clips involving public figures or manipulated footage that influences public perception, all rooted in falsehoods.
A hope lingers that this will mark the pinnacle of AI innovation, that such technologies will plateau due to limits on data or legal regulations. Unfortunately, with existing legislation stalling potential AI regulations for the next decade, optimism in this area appears misguided.
As technology improves without safeguards, the concern is how many policymakers are unwittingly engaging with AI-generated content while remaining oblivious to its implications.
