Absolutely! Here is a Q&A-style summary listing the questions that were asked during the panel—by both panelists (on behalf of the audience) and audience members themselves—along with the answers that were provided. All information here is taken directly from the transcript you provided.
---
## Q&A Summary from the AI Panel
### Q1: For CastMagic and Descript, can you do a landscape video or does it have to be portrait?
**Asked by:** Kelly Stanze
**Answered by:** Janice Person
**Answer:**
Yes, you can do landscape videos. Occasionally, you may need to drag things over a bit, but the system will generally line it up as it thinks it should.
---
### Q2: How much upfront effort does it take to set up CastMagic (for prompts and look/feel)?
**Asked by:** Haley Banwart
**Answered by:** Janice Person
**Answer:**
Setting up the look and feel in CastMagic is relatively quick, as you're feeding in your content and instructing it via prompts. The way you write the prompts and organize them really matters. For example, specifying the target audience (e.g., “new to agriculture”) helps tailor the output appropriately.
---
### Q3: For your custom GPT (Beef Temp Checker), does someone have to manually run it each day or can it be scheduled?
**Asked by:** Kelly Stanze
**Answered by:** Janice Person
**Answered by:** Casey Mills
**Answer:**
You still have to click a button each day to run the check; it isn’t fully automated yet. Most AI tools still require prompting, but the process is simplified to a single button click with defined instructions about sources.
---
### Q4: Are there ways to integrate AI risk-checking/checks into other tools (e.g., social listening, brand reputation management)?
**Asked by:** (Audience, as paraphrased by Kelly Stanze)
**Answered by:** Kelly Stanze
**Answer:**
Yes, most modern AI tools have APIs that allow integration into third-party tools, such as social listening suites or reputation management platforms. It’s also possible to set up automation using coding languages like Python or SQL for custom workflows.
---
### Q5: When you program a custom GPT, do you just enter prompts or is there a more advanced setup?
**Asked by:** (Audience/Panel, paraphrased by Cody)
**Answered by:** Casey Mills
**Answer:**
It’s a lot of refining prompts and instructions until the tool gives consistent, reliable results. You add descriptions, custom instructions, knowledge files (background docs), and keep updating as you use it more. Sometimes, you even get the AI to help you write its own instructions.
---
### Q6: How did your team react to the introduction of the daily risk-checking AI tool?
**Asked by:** Kelly Stanze
**Answered by:** Casey Mills
**Answer:**
They were very open to it, especially after experiencing a painful incident where a tweet blew up unexpectedly. If a tool is at least 85% correct with its risk predictions, that’s already much better than before. The tool is used as part of daily routine content checks.
---
### Q7: (From audience) How do you create custom AI for your clients, and how do you handle data safety/confidentiality?
**Asked by:** Audience member
**Answered by:** Haley Banwart
**Answer:**
Custom GPTs are created using the paid version of ChatGPT, which allows for secure, in-platform data handling that does not push information into the public Internet. Data privacy is a priority, and only select platforms are used after internal vetting.
---
### Q8: (From audience) Is there a way to get unbiased Google search results, not tailored based on search history?
**Asked by:** Audience member (unspecified, but responded to by Kelly Stanze)
**Answered by:** Kelly Stanze
**Answer:**
You can achieve more objective results by using an “Incognito” window (in Chrome) or similar private browsing mode, which ignores your stored browser data and preferences during the search.
---
### Q9: If your website is optimized for AI (LLM) search, do you need to do anything differently for the user/human front-end experience? Or can it be “back-end” only?
**Asked by:** Molly (paraphrased)
**Answered by:** Kelly Stanze
**Answer:**
Good technical optimization and accessibility help both AI and users. Over-optimizing for AI at the expense of human experience is not recommended. Most best practices for search (like proper heading structure, alt text, accessible formatting) are good for both.
---
### Q10: What is a custom GPT and how hard is it to set one up?
**Asked by:** Kelly Stanze (asked the audience for clarification)
**Answered by:** Kelly Stanze
**Answer:**
A custom GPT is a recurring function within ChatGPT: like creating a reusable Word doc for a specialized task. They can be tuned and trained for specific workflows and are pretty easy to create with the paid version of ChatGPT.
---
### Q11: (From audience) What AI tool do you use to check the authenticity or “humanness” of written content?
**Asked by:** Brandi (paraphrased)
**Answered by:** Kelly Stanze
**Answer:**
You can actually use ChatGPT itself: instruct it to check for indicators of AI-authored content. Crafting a clear prompt asking for an authenticity audit is the key.
---
### Q12: (From audience) Have you run into cases where you can’t trust the vocabulary or accuracy of what AI gives back? What do you do about that?
**Asked by:** Audience member, Carol (paraphrased)
**Answered by:** Janice Person, Kelly Stanze, Haley Banwart, Casey Mills
**Answers:**
- You have to provide good context prompts, supply trusted resources, and specify desired vocabulary and tone (e.g., for a “dairy farmer using veterinarian terminology”).
- AI “hallucinates” sometimes, like a child making things up. Always check results for errors or mismatches.
- For client bots, repeated testing, refining, and uploading reference documents (brand standards, key terminology, legal docs) are necessary to keep outputs reliable.
- Paid versions and custom instructions are recommended for quality; free tools are much less dependable.
---
### Q13: (From audience) How do you get organizational buy-in for using AI tools, especially if people are skeptical or concerned?
**Asked by:** Audience member
**Answered by:** Casey Mills
**Answer:**
Be upfront about using AI and clearly communicate the use cases, security measures, and boundaries. Present value through demos and by outlining safeguards. Competitors and detractors are already using these tools, so it’s important not to fall behind. Outline usage policies and ensure transparency.
---
### Q14: What’s the difference between free and paid AI (e.g., ChatGPT) regarding security?
**Asked by:** Janice Person (paraphrased from conversation flow)
**Answered by:** Casey Mills, Kelly Stanze, Janice Person
**Answers:**
- Paid versions of ChatGPT allow you to toggle off data use for model training, keeping your content private.
- Free versions use your data to train the model and your information could “leak” later.
- When handling anything confidential (news, client data, etc.), always use the paid/secure versions and avoid freeloading tools.
---
### Q15: (From audience) Are there any differences in how AI interacts with alternative search engines, like DuckDuckGo?
**Asked by:** Audience member (paraphrased)
**Answered by:** Kelly Stanze
**Answer:**
All major search engines are building in AI layers—some more visibly than others. Even non-mainstream engines (e.g., DuckDuckGo, Baidu) will be incorporating AI modes. Google is the leader, but Bing is most used by current LLMs like ChatGPT and Perplexity.
---
### Q16: What are the differences among various large language model tools—e.g., ChatGPT, Claude, PI?
**Asked by:** Janice Person
**Answered by:** Kelly Stanze
**Answer:**
They’re all slightly different:
- **ChatGPT** is the largest, most feature-rich, and “smartest” due to its vast data set.
- **Claude** (by Anthropic) is more focused on ethics and morality.
- **PI** is tuned for emotional intelligence and human-like conversation.
Choose based on your specific needs: technical, ethical, or conversational.
---
If you’d like a more streamlined FAQ or a grouping by topics (e.g., security, practical use, search/SEO, organizational adoption), let me know!