AI Journalism Truth |635|

Craig S. Smith used to write for WSJ and NYT, now he’s into AI.

After a few AI-only episodes, I was ready to talk to a real person. Craig S. Smith, formerly of the New York Times and now host of the Eye on AI podcast, was perfect.

Here are five key takeaways from Skeptiko 635, and the interview with Craig S. Smith, along with direct quotations from both Craig Smith and Alex Tsakiris:

  • AI as a Truth Engine: The potential of AI to act as a “truth engine” is a significant theme. Craig Smith discusses a project that aimed to compare alternative narratives in media using AI, highlighting the challenges and possibilities in determining truth. He notes, “We wanted to see if we could gather information, gather all of these narratives that exist out in social media and various platforms and compare them to the dominant media and the mainstream and dominant narratives in the mainstream media.”
  • Ethical Considerations and Transparency: The discussion emphasizes the importance of transparency in AI systems over the pursuit of an absolute truth. Craig Smith states, “I think transparency is more important than truth because I think truth is such a slippery concept.” This underscores the need for AI systems to be clear about their processes and limitations.
  • Challenges of AI in Journalism: Smith points out the difficulties in using AI to discern truth due to the vast amount of data required. He explains, “You need an enormous amount of data, not only data on the, more importantly for training. So it would be a huge project, but not one that I think is impossible.”
  • AI’s Role in Logical Analysis: Alex Tsakiris highlights the capability of AI to outperform humans in logical reasoning, suggesting that AI can expose logical flaws in arguments. He asserts, “AI can significantly outperform us just by presenting a solid, logical argument and pointing out someone else’s logical flaws.”
  • Bias and Misinformation: The conversation touches on the issue of bias in AI models, with Smith noting that restrictions on speech affect AI models globally, not just in countries like China. He remarks, “There are restrictions on speech in all of the models that exist out of the US.” This highlights the universal challenge of ensuring unbiased AI outputs.

Transcript: https://docs.google.com/document/d/e/2PACX-1vQBDCB2AQjTjkeCGL-pK5KQ7c_g2PujETlI0TSwC1tk9MhHcR1lU_gxceWre9IFDmn2FlMa0JYChvJH/pub

Youtube: https://youtu.be/fO9Idq5gLLY

Rumble: https://rumble.com/v5ax4ol-craig-s.-smith-used-to-write-for-wsj-and-nyt-now-hes-into-ai.html

[box]

  • More From Skeptiko

  • [/box]