Can you trust AI with your content?
Download as PPTX, PDF2 likes1,846 views
The document discusses the trustworthiness of AI-generated content, revealing that while many agencies are utilizing AI text tools, public trust in such content remains low. Despite AI's capability to rank well and be monetized, there are concerns about data accuracy, originality, and context. The future is anticipated to see increased integration of AI in content creation, yet it highlights the importance of human oversight and expertise.
Can you trust AI with your content?
- 1. Can you trust AI with your content? matbennett.com/brighton2022 @matbennett Mat Bennett Agency Coach, Mentor & Consultant
- 2. Can you trust AI with your content? Words & data: Mat Bennett Images: Midjourney AI @matbennett
- 4. … LinkedIn suggests this @matbennett
- 7. Can they even detect it? @matbennett
- 8. Could you just pump out 1000s of pages of crap and make money on the long-tail? @matbennett
- 10. A show of hands @matbennett
- 12. 65% Already using AI text tools in some capacity @matbennett
- 13. 29% Have used it in a production project @matbennett
- 14. It’s the future… but what is it? @matbennett
- 16. Once upon a time @matbennett
- 17. Once upon a time time time time time time time time time time time time time time time time time time time time time time time time time time time time time time time time time time time time time time time time time time time time time time time time ???? ???? @matbennett
- 18. Once upon a time, there was a poor farmer. He worked hard every day to try and make a living, but it was just too difficult. The farmer had a wife and two children, and they were all struggling. @matbennett
- 19. Once upon a time, there was a computer programmer who was very unhappy with his job. He worked long hours in a cubicle, and his boss was always demanding more work from him. The programmer decided that he'd had enough, and he quit his job. @matbennett
- 20. Once upon a time, a little girl had a dream. She dreamed of a world where she could be anything she wanted to be, where she could do anything she wanted to do, and where she could go anywhere she wanted to go. @matbennett
- 25. @matbennett
- 26. Can they even detect it? @matbennett
- 29. Could someone just pump out thousands of pages of crap and make money monetising the long- tail? @matbennett
- 32. @matbennett
- 33. “Would you trust website content that had been written by AI?” @matbennett
- 34. “Would you trust website content that had been written by AI?” @matbennett
- 35. @matbennett
- 36. @matbennett
- 37. @matbennett
- 38. @matbennett
- 39. @matbennett
- 40. @matbennett
- 41. People don't trust AI content… @matbennett
- 42. … and REALLY don’t trust some of it @matbennett
- 45. How engaging is this article? @matbennett
- 46. How knowledgeable do you think the author is? @matbennett
- 47. Do you trust this content? @matbennett
- 52. The public prefer AI content over cheap human content @matbennett
- 56. Problem: It’s all old data @matbennett
- 57. The current President of the USA is @matbennett
- 58. The current President of the USA is Donald Trump @matbennett
- 59. Problem: Fact & fiction get confused @matbennett
- 61. @matbennett
- 63. “Are you mainly summarizing what others have to say without adding much value?” @matbennett
- 67. What this teaches us @matbennett
- 68. AI text is NOT as good as human writers… @matbennett
- 69. Not everyone will notice @matbennett
- 70. Not everyone will notice (or care) @matbennett
- 72. Topic matters But audience matters more @matbennett
- 73. The gap is narrow @matbennett
- 74. Would I trust AI with MY content? @matbennett
- 75. Would I trust AI with MY content? Unsupervised Today Not if I care @matbennett
- 76. 48% Of agencies expect to be using AI Text tools regularly within 2 years @matbennett
- 77. @matbennett
- 78. This is just the beginning @matbennett
- 79. GPT-3 is already old hat @matbennett
- 80. GPT-2 GPT-3 BILLION BILLION 175 1.5 X177 @matbennett
- 81. GPT-2 GPT-3 GPT-4 BILLION BILLION TRILLION 100 175 1.5 X177 @matbennett
- 82. GPT-2 GPT-3 GPT-4 BILLION BILLION TRILLION 100 175 1.5 X571 X177 @matbennett
- 83. GPT-2 GPT-3 GPT-4 Wu Dao BILLION 175 BILLION TRILLION TRILLION 100 175 1.5 X177 X1000 @matbennett
- 84. “'OK, I think this system is sentient” @matbennett
Editor's Notes
- #23: Unrankable human/unrankable ai.
- #26: The tool isn’t the issue, but what many do with it is
- #29: Their aim isn’t to surface human content. It’s to surface USEFUL content. The source doesn’t matter. The content does
- #33: 853 responses
- #45: 289 responses
- #81: Measured in parameters
- #82: The ways we interact will improve too
- #83: The ways we interact will improve too
- #84: The ways we interact will improve too
- #86: This is happening. AI systems WILL be part of our work. More nuanced than just article gen, but we just don’t quite know how yet. Learn now. Shape that future. Everything you need to get started is as that URL and the next two talks in this room.