Ever since mankind started inventing things, we’ve been obsessed with automation. Just a few days ago, Elon Musk assured me that my car will soon be able to understand my thoughts and feelings and possibly even do my taxes. It appears we will not rest until artificial intelligence is given to everything from a screwdriver to the computers that contain our most precious information.
The human race’s obsession with AI is even more striking, considering there exist few sci-fi stories that paint a rosy picture of the outcome. From Terminator to I, Robot to The Matrix, it seems that nearly every scenario we have envisioned tells us that giving anything more than basic Excel spreadsheet skills to a machine is dangerous. They all end up with the same conclusion: Humans are the problem. There are too many humans. Let’s kill all the humans. End Code.
It’s in this mess that we’ve been introduced to another terror, deep fakes. Over the past few years, the technology needed to map a person’s voice has been steadily improving. Although far from perfect, one can manipulate audio to, say make Barack Obama rap like 2Pac, or Amit Shah say something nice about minorities. This, when coupled with a virtual avatar that looks a lot like the real thing, is a recipe for disaster, and just an idea of what deep fakes are capable of.
And yet, the march to improve AI continues unhindered. Recently, Samsung’s Neon project unveiled a new AI Humanoid Chatbot. Now, if the words AI Humanoid Chatbot do not instil in you an automatic reflex to call your real estate agent and inquire about available post-apocalyptic bunkers in your area, congratulations! You are indeed brave.
The AI chatbot is effectively a virtual avatar that — from early visuals online — looks surprisingly lifelike. Rather than do anything useful, like telling you the weather or reminding you to pay your taxes, these bots are merely there to converse with you and make you think you’re interacting with a real human. Over time, I assume they would begin to behave more lifelike, allowing them to take on the roles of news anchors, virtual receptionists, actors, and teachers.
All sci-fi stories end up with the same conclusion: Humans are the problem.
Now, it should be said that Neon has assured everyone that privacy is their top priority. But then, so did Facebook, and now everyone knows my relationship status is a frown smiley, so any assurance given by tech companies should be viewed with scepticism. Because, even considering the remotest possibility that the company has the best intentions, there is always a human waiting to use technology to prey on other humans. Incidentally, it is this tendency that makes AI’s argument for killing all of us so compelling.
To take a more real-world example, a recent Vice report, investigated how the technology behind deep fakes was used by the BJP in their recently concluded, albeit unsuccessful, election campaign in Delhi. While AAP’s Arvind Kejriwal was out campaigning and giving speeches, the BJP decided to use tech to sidestep the language barrier. According to the report, videos emerged of Delhi BJP President Manoj Tiwari speaking multiple languages, including Haryanvi — a language he has no proficiency in — and English, in a bid to speak the language of each voter.
Shared on WhatsApp, the video went viral and reached over a million people. It contained what’s referred to as a “lip-sync deep fake”, where the person’s voice is dubbed and their lips replaced to make it appear like they are speaking the words.
Deep fakes are not necessarily new. The adult film industry has been pasting celebrity faces on pornstar bodies for years now (a “friend” informs me that these videos are getting more realistic with each passing month). Considering the adult film industry has always been at the forefront of new technology (it’s in their best interest, after all), the quality of deepfakes is only going to keep improving.
With the technology getting better and more accessible, there is the very realistic and worrying prospect of deep fakes being used to spread misinformation. Think revenge porn, or politicians spreading conflicting messages to appeal to different voter groups, or the terrifying fact that as these things get more realistic, misinformation is going to become even easier to spread.
The AI chatbot is effectively a virtual avatar that — from early visuals online — looks surprisingly lifelike.
That, in a world where social media allows us to spread information like never before, where a bored aunt or grandparent is ever ready to blast everyone in their WhatsApp universe with a piece of sensationalist news, is immensely problematic. Imagine a video of Donald Trump declaring war on Iran shows up online, complete with his voice, silly hair and orange skin. Or a video of Christopher Nolan saying he would cast Tiger Shroff in all his upcoming films.
Both these scenarios are likely to cause panic across the world (different worlds of course), damage that’ll take years to be undone.
Basically, we’re reaching a tipping point, where the intention to spread misinformation is just about coinciding with the technology to make it believable and an unprecedented ability to spread it at zero cost. For now, we can rest uneasy knowing that the technology is still a long way from being truly convincing. So, enjoy your chat with a virtual Beyonce. Engage in healthy banter with a Rajdeep Sardesai bot. But for the sake of sanity, if Beyonce starts asking for your mother’s maiden name or the city in which you were born, exercise some caution.
We can’t go about blaming AI if we’re being just plain stupid.
This article was first published in Arre
Stay updated with all the insights.
Navigate news, 1 email day.
Subscribe to Qrius