By Brian Redmond, PhD and Bing AI
This is the second entry in a series of posts about generative AI and what it means for your writing. This entry and the previous one in the series focus primarily on fact checking AI. That is why it I had it write blog entries about me. That way it was very easy for me to fact check, but the same principles apply to any information that it would generate.
In this entry I’m going to expand my “fact check” annotations a bit to include more general analysis. The reason being that I put Bing AI in its “more creative” mode option. This mode is basically ChatGPT 4 where the last entry was in “more precise” mode (basically ChatGPT 3.5). You’ll see that things react a bit differently. For example, with the same prompt, of “Please write a blog about Brian Redmond who teaches at Penn State” you’’ see some significant differences. One of which is that the “more precise” version generated 282 words where the more creative generated 465 words; almost 65% larger without any instruction to do so.
To help avoid confusion over who wrote what, what I write will appear in bolded or normal font. And what Bing AI writes will be in italics.
Meet Brian Redmond, an Online Teaching Professor and Director of Organizational Leadership and Labor & Human Resources Programs at Penn State
If you are interested in learning about leadership, psychology, and online education, you might want to check out the courses taught by Brian Redmond at Penn State World Campus. Brian Redmond is an online teaching professor and the director of the organizational leadership (OLEAD) and labor and human resources (LHR) programs at Penn State. He has been teaching online for over 15 years and has been involved in many innovative initiatives to enhance the online learning experience for students and faculty.
Brian’s fact check and analysis: This is all true, and unlike the previous entry, this information is more up to date. The “if you are interested” phrase is intriguing to me, as the AI is trying to interject more emotion into it’s writing that it didn’t have in the last entry to make itself seem more like human writing.
Brian Redmond has a PhD in industrial and organizational psychology from the Graduate School & University Center/Baruch College. His research interests include leadership, motivation, industrial and organizational psychology, and online education. He is also a founding member of the Online Leadership Development Institute (OLDI), which helps assess and develop World Campus students’ leadership competencies1.
Brian Redmond teaches courses in leadership, such as introduction, communication skills, global, among others. He also co-created the psychology of leadership master’s degree and co-founded and taught in an inclusive leadership certificate. He is passionate about helping students develop their leadership potential and apply their skills in various contexts.
In addition to teaching, Brian Redmond also serves in many leadership roles at the department, college, and university levels. He is the immediate past chair for the Commission for Adult Learners, which advocates for the needs and interests of adult learners at Penn State. He also co-founded and co-chaired the Coordinators of Online Learning and Teaching, which is a network of faculty and staff who support online education at Penn State. He has also been involved in strategic planning, curriculum revision, faculty development, student engagement, and conflict resolution.
Some of the achievements that Brian Redmond is proud of include launching the first psychology club in the nation for online students2, experimenting with mobile robots to help give online students access to campus resources2, developing virtual reality simulations to help students practice leadership skills2, and planning two World Campus Faculty Convocations2.
Brian’s fact check and analysis: The four paragraphs above are all true. Compared to the last entry, “more creative” does a better job adding more details as well as synthesizing the ideas into a more coherent story than the “more precise” option. This makes it harder to distinguish from good human writing. That said, it far exceeds new or poor human writers, so that can be a subtle hint in detecting AI generated content if you know the writer’s natural writing ability.
Brian Redmond is not only a leader, but also a learner. He is always looking for new ways to improve his teaching and his programs. He values feedback from his students and colleagues and strives to create a positive and collaborative learning environment. He is also open to new challenges and opportunities that can help him grow as a professional.
If you want to learn more about Brian Redmond, you can visit his website1, read his CV3, or contact him at [old email address redacted]4. You can also enroll in one of his courses or programs at Penn State World Campus and benefit from his expertise and experience. You will not regret it!
Brian’s fact check and analysis: These last two paragraphs are also all true where it states facts. However, like in the last entry, as Bing AI is drawing a conclusion it tries to invoke an emotional appeal to make the post more appealing to you as the reader. It is a bit weird to me that it is speculating about my intentions. It is not entirely wrong, but it clearly has never met me and the speculations are vague enough to be true about any professional who is progressing through their careers. That is another subtle hint at AI writing; conclusions that lack a certain depth; they aren’t wrong, but rather generic enough to be true about any similar situation, not just the specific circumstance at hand. Something else that I have picked up on is that it does tend to get a bit salesy. For example see line about taking one of my classes rather than drawing a conclusion about my expertise or something more substantive.
Learn more:
Brian’s fact check: These are the same legitimate sites that “more precise mode” pulled up in the last entry. So just like then, one of the citations is a dead link from a site I took down. Which is another subtle indicator of AI writing; inaccurate citations.
Brian’s overall all analysis and conclusions for your writing
AI writing can be hard to detect. The clues are subtle. Here is a quick summary of the ones that I have compiled from the first two entries in this series on AI writing:
- Inaccurate citations
- Inconsistencies in language
- Facts may be outdated
- Conclusions tend to be true but generic/lacking depth
- It’s tone can be very different from the human that made the prompt
- Similarly the writing style can be significantly different from the human it is supposed to be representing
In comparing a ChatGPT 3.5 and ChatGPT 4.0 based generative technologies, there is no question that ChatGPT 4.0 is much more effective in mimicking human writing. Currently, most people have access to ChatGPT 3.5, with a growing number having access to 4.0. So detecting the difference between human writing and generative technology is going to get more difficult in the coming months and years.
This isn’t necessarily a bad thing. We need to start thinking about writing a bit differently. The major weakness for humans in generating content is writer’s block. Generative technologies don’t have that issue. If people can generate brainstorm ideas or rough drafts using these tools, we can overcome the writers block that keeps people from sharing their ideas. That said, there are a few ethical concerns with that. Who get credit for the writing? Will we generate so much content that there will be nothing left to be said? I’m going to explore these ideas and some others in more depth later in this series.
One thought on “AI Series #2: Generative AI fact checking Part 2 and a bit more analysis”