How to use generative AI for Content SEO and PR — Kristin M Tynski // Fractl

Kristin Tynski, Co-Founder and SVP Creative at Fractl, talks about generative AI for SEO and PR. Generative language models like ChatGPT are essentially like having an AI-powered personal assistant with infinite knowledge at your fingertips. In PR, this means that with the right prompts and information, you can create attention-grabbing content that's personalized and timely with just a few keystrokes, making your PR efforts more efficient and effective. Today, Kristin discusses how to use generative AI for content SEO, and PR.
About the speaker

Kristin Tynski

Fractl

 - Fractl

Kristin is Co-Founder and SVP Creative at Fractl

Show Notes

  • 02:17
    Best practices for using ChatGPT in PR
    There are various potential applications of ChatGPT in PR, but it is recommended to utilize it as an inspiration or refinement tool for human-edited work. It is essential to avoid solely relying on generated text from ChatGPT without human review, as it may contain inaccurate information.
  • 04:01
    Tips for injecting human judgment in the QA process
    Human judgment is crucial in the final stage of the QA process. However, with prompt chaining, you can use ChatGPT to do some of the legwork in quality control by iteratively refining the output based on defined criteria or rubrics.
  • 06:12
    Leveraging ChatGPT to understand journalist beats for effective PR pitching
    In order to effectively do PR, it is important to understand the appropriate reporter beats to pitch to. By feeding your article or article title to ChatGPT, you can determine the potential beat angles that may be appropriate for your content.
  • 08:13
    Using ChatGPT to suggest pitch angles from written content
    This process involves inputting a full article or content and asking the model to pull out the most interesting and newsworthy takeaways or facts. This can be done iteratively, and the results can be sorted by interestingness, newsworthiness, or relevant beats.
  • 09:10
    ChatGPT assistance in writing sample pitches
    ChatGPT can assist in writing sample pitches for multiple journalists across different beats, using an iterative process to tailor the pitches to each individual. While this can save time, it is important to have a human review the pitches before sending them.
  • 09:53
    Using ChatGPT for pitch evaluation and editing
    ChatGPT can evaluate and improve pitches to make them attention-grabbing, timely, and personalized to the journalist. It can also suggest interesting subject lines, which are crucial for open rates, make the pitch concise, add calls to action, and format for readability.
  • 11:58
    Using ChatGPT for press releases
    ChatGPT can generate well-formatted press releases based on limited information, such as company and contact details. Fractl has future plans to incorporate prompt chaining and external data sources to better personalize drafts for specific journalists and their audiences.
  • 14:24
    Learning to engage with generative language models
    Navigating generative language models like ChatGPT requires learning to blend your knowledge with its extensive knowledge. Its a process of discovering what you don't know and learning how to interact with the model to get the most out of it.
  • 16:06
    Biases and safeguards in generative language models like ChatGPT
    Generative language models have biases built into them since they were trained on human data. Open AI has safeguards at the prompt level to modify their behavior, however, people have figured out how to jailbreak these prompts and push the limits of the base model.

Quotes

  • "ChatGPT is prone to making things up from time to time. But that doesn't limit the potential of it as long as you have a human in the loop as part of your process." -Kristin Tynski, Co-Founder, Fractl

  • "If you run and rerun a prompt three to six times, and get the same answer each time, you can be pretty sure that's correct. Of course, you should still verify it." -Kristin Tynski, Co-Founder, Fractl

  • "The overarching start to doing PR is understanding what audiences, reporter, or journalist beats are appropriate for pitching to." -Kristin Tynski, Co-Founder, Fractl

  • "The open rates on pitching are extremely dependent on how interesting your subject line is." -Kristin Tynski, Co-Founder, Fractl

  • "The biggest learning curve with working with generative language models like GPT-3 is figuring out how to engage with them. It's almost like getting to know a person or an alien intelligence." -Kristin Tynski, Co-Founder, Fractl

  • "Before you ever have any conversation with chatGPT, Open AI has an initial prompt that you don't see that is modifying its behavior beyond what the base model itself would do if that weren't there." -Kristin Tynski, Co-Founder, Fractl

About the speaker

Kristin Tynski

Fractl

 - Fractl

Kristin is Co-Founder and SVP Creative at Fractl

Up Next: