AI, deepfakes, and the mess we are getting ourselves into

Has AI gotten out of control? We live in a world where anything from a YouTube meme to a voice call from the President of the United States can be faked. AI can change the course of elections, scam people out of millions, or even trick the entire internet. AI is becoming so realistic that many people can’t tell the difference between human and AI content. But is this level of AI content and use good for the online scene as a whole?

There are so many problems with an AI primary internet environment. The first is deepfakes and the legitimacy of claims made by generative AI. During the 2024 election cycle, there were many Deepfakes. There were videos of Taylor Swift endorsing Donald Trump.  There was also a voicemail telling democrats to stay home on the day of the primaries and not vote for Biden. Neither was real, but AI can replicate celebrities and politicians scarily accurately because of the amount of data about these public figures. Both incidents have caused great harm.

Confusion isn’t the only problem, though. This year, a finance worker got on a call with their boss, who told them to send 25 million US dollars to them from their organization. The finance worker did just that. But later that day, when they checked with the corporate offices, they learned they had just fallen for a deepfake scam. They had just been scammed out of 25 million dollars. 25 million dollars.

How can you avoid scams and deepfakes? Look for consistency mistakes. Does the lighting stay consistent? Do wrinkles and other facial features look natural? AI often makes inconsistent errors with small details, and that can give it away.

Another problem with AI is the amount of information on the internet and if AI’s responses collapse. Around 57% of online content has been generated by AI or translated by it. As I said earlier, this poses a problem with legitimacy as well as to the AI itself. When AI learns from its own material, it erodes from accurate information to nonsense. In a study, scientists made a pre-made wiki with only AI-generated answers on it. It was updated whenever the AI made a new answer; they would ask questions and see how fast the answers degraded. By the 9th cycle of questions, a discussion on church steeples turned into a thesis on jack-tailed rabbit colors. It’s like playing a game of telephone, but eliminating random letters every time. AI will try to create a world that falls within its answers instead of an accurate one. We are already at the point where AI content outnumbers human content, but what will happen once the internet is 90% AI or 99%? If AI starts saying insane stuff, will humanity be able to discern the truth, or are we no smarter than the online media we consume?

So what is the big picture? AI has already started spinning stories and content on social media platforms. People have been tricked out of millions. We can no longer trust whether the videos we see are real. AI has taken over the internet, and it will only grow its control. AI has its upsides and its downsides. The final question is, DO YOU WANT AI IN YOUR LIFE?