Magical Pistachios & AI Ethics | Han Liliana Q | TEDxYouth@NIS

Key Concepts:

• Deepfakes: AI-generated content (image, video, audio) designed to mimic real people or scenes.
• AI Ethics: Moral principles governing the development and use of artificial intelligence.
• Plagiarism: Presenting someone else's work or ideas as one's own.
• Critical Thinking: Analyzing information objectively and forming a reasoned judgment.
• Information Credibility: The trustworthiness and reliability of information sources.

1. The Power and Peril of Deepfakes

• Deepfakes, defined by the Alan Turing Institute, are AI-generated content designed to mimic real people or scenes.
• The speaker introduces the concept with a fictional example of Joe Biden talking about a "magical pistachio" to illustrate the absurdity and potential believability of deepfakes.
• AI manipulates people into believing things said by credible figures, influencing their thoughts.
• The speaker emphasizes the prevalence of misinformation disguised in social media, news, and videos.

2. Personal Experience with AI and Plagiarism

• The speaker shares a personal experience where their essay was falsely flagged as 96% plagiarized by Turn It In, an AI-powered plagiarism detection tool.
• This incident motivated the speaker to delve deeper into AI and its potential for misuse.
• The speaker took an artificial intelligence course on Global Online Academy.

3. Real-World Examples of AI Misuse

• Legal Case (February 24th, 2025): Three lawyers were penalized for using AI-generated fake cases in court; eight out of nine cases were fabricated by AI. This highlights the potential for AI to undermine the legal system.
• Financial Fraud (2024): A finance worker approved a $25.6 million transaction after a deepfake video call with their CEO, demonstrating the potential for significant financial losses. The worker had doubts but the video call was so real that he had to put the doubts aside.

4. Statistics on Deepfake Awareness and Impact

• One in four leaders are unfamiliar with deepfakes.
• 31% of people underestimate deepfake fraud risk.
• One in ten executives have already faced deepfake threats.
• There was a 10-fold increase in deepfakes detected globally across all industries in 2023.
• The increasing accessibility of AI makes it easier to fabricate content.

5. Strategies for Identifying and Combating Deepfakes: "Read, Look, Listen"

• The speaker presents a three-part strategy to protect against manipulation by AI-generated content: Read, Look, and Listen.

  • Read:
    • AI text is present on various platforms.
    • Example: AI chatting on a platform for selling secondhand products fabricated false descriptions of a teapot.
    • Look for weird words, phrases, and sentences, inconsistencies in style, and lack of specific details.
    • Check replying speed; AI responses are often instantaneous.
    • Always check for the credibility of the information provided.
  • Look:
    • AI has grown to encompass arts and media.
    • Identify AI images online.
    • Example: The speaker's grandma was convinced an AI-generated image was real.
    • Look for lack of detail in the background, surreal elements (e.g., feet that look like hands, extra fingers), and overly high saturations.
    • The speaker presents a visual example of two images and asks the audience to identify the AI-generated one.
  • Listen:
    • Deepfakes are becoming harder to identify.
    • Upon taking notice, look for sound and visual inconsistencies, differences in noise patterns, color and sound differences between edited and unedited portions of the video, and sync issues between voice and mouth movement, also between that and background movement.

6. Conclusion

• By being aware of the potential existence of AI in all information, we can mitigate the effects of propaganda.
• Even small actions in identifying deepfakes contribute to a larger effort.
• The speaker reiterates the importance of reading, looking, and listening before reaching a conclusion.
• The final question to ask is: "Does this look credible?"

Key Quotes:

• "Follow me and I'll lead you to safety." (Fictional quote from a magical pistachio, used to illustrate the absurdity of deepfakes)
• "Defined by the Alan Turing Institute as an AI generated image, video or piece of audio content that is designed to mimic a real life person or scene." (Definition of deepfake)
• "Read, look, and listen before reaching a conclusion." (Call to action for combating deepfakes)
• "Does this look credible?" (Final question to ask when evaluating information)

Technical Terms and Concepts:

• Deepfake: AI-generated content designed to mimic real people or scenes.
• Alan Turing Institute: A leading institute for data science and artificial intelligence.
• Turn It In: An AI-powered plagiarism detection tool.
• Global Online Academy: An online learning platform.
• AI Ethics: Moral principles governing the development and use of artificial intelligence.
• Neural Network: A computational model inspired by the structure and function of the human brain.

Logical Connections:

• The speaker begins with an absurd example to introduce the concept of deepfakes and their potential for believability.
• The personal experience with plagiarism serves as a catalyst for exploring the broader implications of AI misuse.
• Real-world examples and statistics demonstrate the tangible risks and growing prevalence of deepfakes.
• The "Read, Look, Listen" strategy provides actionable steps for individuals to protect themselves from manipulation.
• The conclusion emphasizes the importance of awareness and critical thinking in combating the spread of misinformation.

Synthesis/Conclusion:

The video effectively highlights the growing threat of deepfakes and AI-generated misinformation. By combining personal anecdotes, real-world examples, and practical strategies, the speaker provides a compelling case for increased awareness and critical thinking in the digital age. The "Read, Look, Listen" framework offers a valuable tool for individuals to navigate the complex information landscape and avoid being manipulated by artificial apparitions. The main takeaway is that vigilance and a healthy dose of skepticism are essential in discerning truth from fiction in an increasingly AI-driven world.