Seeing Is No Longer Believing: Understanding the Rise of AI Deepfakes

Summary

For most of our lives, we have relied on a simple rule: if you see a person speaking with your own eyes and hear them with your own ears, you can be reasonably sure that what you are witnessing is real. However, we have entered a new era where digital technology can now convincingly mimic human beings.

Watch the video below featuring Jensen Huang, the founder and CEO of Nvidia, the largest and one of the most influential companies in the world today. They design the specialised computer chips that power Artificial Intelligence (AI). In this video, Jensen Huang appears to be giving a lengthy, twenty minute presentation on a stage, dressed in his familiar leather jacket, discussing the future of technology.

The footage is remarkably polished. He looks like himself, moves like himself, and speaks with his usual tone and authority. But there is a catch: the video is entirely synthetic. It is what is known as a “deepfake”, a video created by a computer to make it appear as though someone is saying or doing something they never actually did.

What Was the Message of the Video?

Before looking at how it was made, it is worth noting what the “digital” Jensen Huang was actually saying. The video is set in the near future, specifically the year 2026.

The core message is one of extreme urgency. The AI generated speaker explains that we are currently in the middle of a new industrial revolution. He describes a world where AI is no longer just a curiosity, but a tool that is fundamentally changing every part of our lives. He mentions that computers are now discovering new life saving medicines in a fraction of the time it used to take and that robots are becoming capable of performing complex tasks in homes and factories.

His central point is that the pace of change is accelerating so quickly that individuals and businesses must adapt immediately. His recurring advice in the speech is to “run, don’t walk” toward this new future, warning that those who wait will be left behind by history.

How Was This Video Created?

To the casual observer, creating such a video might seem like a feat of Hollywood level magic. In reality, it was likely put together using three distinct layers of AI technology.

1. Generating the Script The creators did not need to be expert speechwriters. They likely used a Large Language Model (similar to ChatGPT). By giving the computer a prompt such as, “Write a twenty minute speech about the future of AI in the style of Jensen Huang,” the computer can instantly generate a script that uses his specific vocabulary and professional logic.

2. Cloning the Voice To make the speech sound authentic, the creators used “voice cloning” software. By providing the computer with just a few minutes of high-quality audio from Jensen Huang’s previous real world interviews, the AI analysed his pitch, his accent, and the unique rhythms of his speech. Once the AI “learnt” his voice, it could read the new script aloud, sounding almost indistinguishable from the real person.

3. Animating the Visuals The final step is the most impressive. Using a technique called “generative lip syncing,” the creators took an existing image or video of Jensen Huang. The AI then manipulated the pixels in the video, specifically around the mouth and jaw, to ensure the lip movements matched the new audio perfectly. This creates the illusion that he is physically speaking words that were actually generated by a computer.

Why This Matters to Us

While this specific video was created to demonstrate the power of technology, the implications are significant for all of us. As these tools become more accessible, it will become increasingly difficult to tell the difference between a genuine news report and a manufactured one.

Scammers and bad actors can use this technology to create convincing videos of public figures, or even family members, to spread misinformation or attempt to deceive people.

How can you protect yourself? Fortunately, the technology is not yet perfect. If you look closely at these videos, you can often spot “tells”:

• The Eyes: The AI often struggles to replicate natural blinking or the way light reflects in a human eye.
• The Mouth: Sometimes the edges of the mouth can look slightly blurry or “glitchy” when the person speaks quickly.
• The Emotion: While the voice sounds real, it often lacks the subtle emotional shifts and imperfections of a living, breathing human.

The goal is not to become cynical or fearful, but to develop a healthy sense of scepticism. As we move forward, we must remember that in the digital world, seeing is no longer proof of the truth. We must learn to verify information from multiple trusted sources before we accept it as fact.

References

• Gemini
• Video