Should we be scared of ChatGPT?
Dr. Caroline Ritter, MBA, DBA
Digital Transformation | IT Strategy | Global Service Delivery | Agile | DevOps | IT Application Delivery | Risk Management | Cost Control | Operational Efficiencies/Automation | Data Privacy/Security
I'm not sure about you, but Chat GPT seems to keep coming up in many different situations and conversations lately. A few weeks ago, another parent at a college tour asked the admissions panel what the college was doing to prevent the students from using ChatGPT to write their essays. Their response was that they were looking at how to restrict its use. A couple of days later, I saw an article about the gymnast and influencer Olivia Dunne having posted about using caktus.ai (a ChatGPT-like tool for students), and LSU quickly responded with warnings of academic misconduct*.?
However, when my 16 yo daughter was working on a school assignment to create a TikTok-like video depicting a day in the life of John Maynard Keynes (the English economist), my first suggestion was to recommend that she ask ChatGPT what his day would have been like. From there, she got a very simple hour-by-hour suggestion that she then had to validate and research, including making sure that the timing of publications and discussions she wanted to include were correct for the timeline and college. Ultimately she used multiple other tools for the assignment, including a text-to-speech generator - adopting an English male voice to narrate the video, editing software, music clips, subtitles, etc., to create something in a few hours that would have taken days or weeks a few years ago. I was really impressed with what she achieved! (and a little jealous.)
When I was studying for my doctorate a few years ago and had an unfamiliar topic, I started with Youtube videos to understand the concept, then I transitioned to published articles in HBR or similar, and soon I was able to understand the peer-reviewed papers that I needed to use as references. If only these tools had been available when I was at school!?
Do we need to ban these tools?
So rather than banning these tools, why not change the assignment to assume the tools are going to be used? As it turns out, my mentor and professor, Michelle Preiksaitis, JD, PhD , is doing just that! As she pointed out to me, ChatGPT does not have reference data after 2021, nor does it provide references for the responses it generates (although the new GPT-4 is reported to do this). So by requesting all input provided by ChatGPT have a peer-reviewed reference and a threshold for the number of required references prior to 2021, it's likely Dr. P's students are able to produce higher quality work more quickly, using today's tools. Did schools ban encyclopedias and demand that students use the original books? How about calculators? Google Search? I can't imagine how much lower the quality of my research would have been without the internet. At the time of writing, there are tools entering the scene that aim to detect the use of ChatGPT in text (e.g., ZeroGPT), and ChatGPT is looking at ways to watermark its output, so the use of the tools would exclude a full "cut & paste."
领英推荐
It seems to me that the conclusion should be - use the tools… just change the rules. What do you think?
Part 1 of 2.
Disclaimer: the opinions expressed in this article are my own and do not represent those of my employers, past or present.
Retired Senior Developer at Sargent & Lundy, LLC
1 年Chat GPT should be in the schools at all levels. The teachers will actually taught how to teach the students how to work with your TP because they'll be working with it in in the future and chat keeps you will continue to evolve faster in the curriculum so the teacher is going to be in the good position they're older and more experience but the kids going to learn faster and now it's either in grammar school they won't start working till they're 21 and said TPT will have changed the face of most jobs jobs that can be automated will be but they're always be jobs for somebody to do something but these other folks are talking about you know they don't really know software like I do I mean I don't know how fast you can change when I started 73 then Howard cards and about every decade things change a lot and by 1993 things are networked by 2000 big day bases were there 2010 and multi multiple instruction multiple data computers the Google and Amazon have millions of computers and they're doing well AI is will continue to evolve but the danger is there needs to be a human in the loop to cut it off and nobody's sharp enough is they going to have some kind of manual cut off because the chat tbts and it may decide
Managing Director at Accenture
1 年Hello Caroline! Yes....Intriguing question. Chat GPT is certainly a robust resource. Although, I think it can make a ton of mistakes as well. I was using it the other day asking about some sports trivia, and the GPT didn't even know what year it was. A double-edged sword for sure. Hope all is well with you!
Medical Scibe at Massachusetts Eye and Ear
1 年Amazing article!
TD Bank | Technology Executive | Board of Directors | Trustee | Transformation Leader
1 年Great article! Well done!
Instructor & Grad Advisor | Business Scholar-Practioner | Author | Telecom Pro | AI Ethics, Tech Adoption | Content Strategist | Passionate about empowering individuals & orgs to reach their full potential.
1 年As an avid student of ChatGPT and AI in general, your article caught my eye. I recently discussed using ChatGPT responsibly when creating outlines, generating ideas, or summarizing information with my graduate students. I was sharing with them that some colleges and universities are at a crossroads with ChatGPT, either using a closed policy that bans students from using it or having an open policy that encourages them to use it responsibly (guard rails). I totally agree with Dr. P that Chat GPT doesn't provide the actual reference source of the information it generates and may generate bogus citations, especially peer-reviewed articles. Part of using ChatGPT responsibly is validating the information it produces or providing ChatGTP with the proper context (prompt engineering). I've seen ChatGPT produce some amazing results and some unexpected results. The key to getting good results with ChatGPT is the quality of the prompts given. Give it vague prompts, and it will produce vague responses. Give it a well-engineered prompt, and it will likely produce good results. Good article; thank you for sharing your experience.