AI technology can now create a copycat with only three-minutes of footage. But is it any good? You be the judge.

Artificial Intelligence has long-populated science fiction. The raw dystopia of Ridley Scott’s Blade Runner famously features a rugged Harrison Ford navigating a world where humans and their ‘replicant’ counterparts are eerily indistinguishable. But as the credits roll, you can happily turn off the TV with a remote control that won’t try to kill you.

But in recent years, AI replication is an increasingly real-world question, with audio and video generation platforms gaining currency. Intrigued, we staged a Bruizer test to see if we could create convincing AI versions of our team.

We commandeered an edit suite to set up an interview space and capture some talking heads. Our control text: a local newspaper article, which provided the basis needed by the AI-platform to generate a copycat. Alongside, we captured a natural piece to camera to compare. In the interest of a fair test, we wanted to generate exactly the same piece using the AI-platform, with only some minor changes, so set up a direct comparison between the real Bruizerite and the AI version.

The answer is, probably, yes. Although undoubtedly lifelike, there’s a distinctly inhuman quality to our techy twin. Firstly, there’s an unnaturalness to the facial movements. In some cases, the movement is isolated to specific areas, such as the mouth and eyes – depending on the natural habits the AI has picked up on – but it largely feels irregular or exaggerated.

There’s also a leaning towards overly regulated rhythms of speech, which teeters on the robotic at points. Even for the original Bruizers who don’t boast a strong regional accent, the metronomic syllables of their AI are too uniform for even a well-spoken voice.

Admittedly the gestures are uncanny in places and very convincing in how they mimic our idiosyncrasies. The occasional head tilt or nod generally has good alignment to where we’d naturally place them, introducing a question or ending a sentence.

But for all the accuracy, the overall AI character is disinterested, disengaged. We hope it doesn’t sound too big-headed, but there’s no question the real Bruizers are far more enthusiastic than our AI counterparts who, it appears, would really rather be somewhere else.

Our weekday cloning session was only ever intended as controlled experiment, but it can’t help but pose a larger question: what are the real-world implications of this technology for the filmmaking industry?

It’s possible we’re at the beginning of a talking head revolution. A company’s CEO could cut down their interview time, needing only three minutes of content for a ten-minute end of year message. Perhaps we can say goodbye to the laborious process of multiple takes, trying to perfect the delivery of a lengthy script.

If it’s not generation, the same AI-platform we used to produce the fake Bruizers also has the ability to transpose into different languages. Multi-national companies could have the opportunity to improve accessibility and communicate direct to their employees across the world in their native language. Although therein lies another fault, that the level of fluency and pronunciation achieved by AI is unrealistically perfect for non-native speakers.

On the other hand, even native speakers make mistakes so a more subtle application of the technology might be the ability to correct grammatical slips or mispronunciations in post-production, rather than needing to organise a reshoot.

And it’s not all hypothetical: an example of an already established AI application is eyeline correction on facetime calls. Where online communication is considered cold and disconnected, these small changes are already creeping in as an effort to make the experience more personal.

The question is, can it ever compare to real human contact? If our experiences of the COVID-19 pandemic taught us anything, it’s that we can coexist through the internet, learn via pre-recorded lectures and collaborate over video call. But do we want to?

Whatever the application, at the core of the technical debate is a deeper, existential one: how far do we distinguish AI from the real person behind it? Is an AI-being separate, or an extension of ourselves?

The inherent fear that surrounds AI as a concept is derived from such questions. The underlying cautionary tale of much science fiction plays into our fear that one day there will be an intelligent being that can outsmart, outmode and outlive us.

If you really want to challenge your philosophical thinking, consider the paperclip theory. To paraphrase Nick Bostrom’s famous thought-experiment, an AI is tasked with making paperclips. Having used up all obvious materials for crafting stationery, the AI pursues further resources and eventually burns through all the raw materials on earth. The AI then progresses to a new planet in search of more paperclip fodder, the ultimate result being that a menial task spells the destruction of the universe and us mere mortals within it

So, more than ever, it’ll be important to prove the realness of things and authenticate the messages we put out into the world. A company will be measured by how its brand is set apart from the figment of technology. As a result, we could see an increase in live broadcasts, to engage employees in real time, with undeniably real people. Or a move away from generic messaging in recruitment or corporate campaigns, driven instead by existing colleagues interacting with the workplace and emphasising action over spoken word.

It’s an old adage among creatives but true: ‘show, don’t tell.’ In a world made bland by AI-content, the talent of videographers will lay in capturing what can’t be imitated: the character and connection of real people, the organic ethos of a company, the energy of a brand. It’ll be more important that ever to partner with creative agencies who can develop bespoke styles and create content that doesn’t just tell you something’s true, but makes you feel it too.