Chatterbots

Breaking Bogart (05)

Chat GPT logo

Welcome back to another week of Breaking Bogart. I will be writing this week about my experience with generative artificial intelligence (Gen. A.I.) and interacting with a chat bot. Specifically, I interacted with Open A.I.'s model ChatGPT4o. I have historically been very critical of the usage and ethics associated with using Gen. A.I. This does not mean I am ignorant to the benefits of utilizing or applying this technology. More so, I believe that it is now more important to ask if and how we should be pushing Gen. A.I. forward in its capabilities and applications.

Humans are social and emotional creatures. This is not an inherently negative thing nor does it invalidate our needs or how we live. On the contrary, we use emotions to assign value to our experiences. These experiences shape how we view the world and make decisions. Therefore our emotions are the basis of how we perceive fact and reason. 

Connection

My thesis on the matter of A.I. usage has always been the following: Gen A.I. is a tool to be used or misused. When the tool is pushed past being a tool to the point of acting as a human substitution, there is a vital and crucial component to human experience that is lost. I believe this concept goes so deep as to be involved with the human purpose for living. We go back to the fact that we are social creatures and crave interaction with others who are sharing the same or similar experiences as us. The fact that we are all trying to make our way through life provides an inherent and shared struggle binding all of humanity together. I do not believe that A.I. ever will, nor should it try/succeed to replace that aspect of life. 

For the purposes of discussion, let's consider this example: Gen A.I. has reached a hypothetical point where it is now capable of teaching a college-level course which covers a sufficiently complex, nuanced, and applied topic. It is able to lead and manage students, deliver information, proctor exams, adjust its approach according to student feedback, and otherwise fulfill any and all roles a teacher needs to fill. In this example, students will never form a human or emotional connection with this form of intelligence. Even if this Gen A.I. is now capable of perfectly mimicking human tendencies to manipulate a student's emotions to simulate the human connection, (as multiple quite terrifying models marketed to young adults are capable of doing) it does not suffer through the human condition. All of this is to say this: humans create for the benefit of humans. If we are to be capable of replacing such an essential role as "the teacher" with an artificial intelligence, does that serve humanity? I do not think it does. Therefore this intention must not be pursued as it is being pursued now.

That will conclude my opinions on the ethics of Gen A.I. I believe it is a massively important discussion that should always be had prior to losing our heads over what it could possibly achieve. (This is not even getting into the power consumption and socioeconomic dilemmas that Gen A.I. has and will continue to worsen).

As for my experience interacting with Gen A.I., here it is. Below in quotes is my prompt to Chat GPT4o. (The information was cross referenced with resources I had already located and saved and then used towards the completion of my individual technology tool research paper).

"Could you please now research on the web all of the available resources a ThingLink user could access for training and as a general resource? Please ensure resources are either originating from the ThingLink team or are approved by them. List your findings in a bulleted list with hyperlinks to the main webpage included as well."

What the chat bot responded with:

***

"Here are some valuable resources from ThingLink for users seeking training and support:

• ThingLink Professional Development Resources: This page includes various professional development resources, such as the ThingLink Certified Creator course. This self-paced course covers the basics of ThingLink as well as advanced options, including visual design and accessibility tips.

• ThingLink Learning Center: The Learning Center offers a comprehensive range of tutorials, guides, and instructional videos to help users get started with ThingLink and explore advanced features.

• ThingLink Education YouTube Channel: The official YouTube channel provides additional guides, interviews with educators, and webinar recordings, offering visual and practical support for users.

• ThingLink Education Blog: The blog contains a wealth of articles on best practices, case studies, and innovative ways educators are using ThingLink. It also features information about upcoming webinars and events.

• ThingLink Help Center: The Help Center is an extensive resource where users can find answers to frequently asked questions, troubleshooting tips, and detailed articles on how to use various ThingLink features.

• ThingLink Certified Educator Program: This program offers certification for educators who want to demonstrate their expertise in using ThingLink to create interactive learning experiences.

These resources, provided and approved by ThingLink, ensure that users have access to comprehensive support and training, enhancing their ability to utilize ThingLink effectively in educational settings."

***

Reflection

I have summarized my thoughts on its response and general performance below:

• The bot did ok. It used the same website and webpage/resource as two separate resources and just described it in a different way. I understand that Gen A.I. does not produce perfect responses first try, but merely reusing an answer twice seems a bit excessive for the latest and greatest Chat GPT model.
• It also hallucinated a url to a youtube channel that does not exist. I pointed this out and asked it to be corrected but the bot hallucinated another, different url to a youtube channel that also did not exist. I had to provide the exact url before the hyperlink was corrected. Again, more underperformance that surprised me for being the latest model.
• The bot has a tendency to reuse a lot of the same points when asked different questions. Even when explicitly asked to generate new ideas or assertions.
• The bot always produces a semi-vague and roundabout way of forming its sentences and presenting ideas.
• The bot does an incredible job at receiving directions geared towards the length and formatting of its response.
• The bot has incredible potential for saving time in mundane tasks or even combing over large groups of data or data location information.

Something I learned from this experience is that no matter how specific and well written your prompt to a chat bot or Gen A.I. may be, you may have to repeat yourself several times or change how you word said prompt in order to receive a satisfactory answer. This was surprising to me as from what I understood, Chat GPT matches its style and complexity of conversation with the author's perceived level. I suppose I did not consider that this does not mean that the truth or validity of the response will also rise to match.

I will send off with this thought: I believe there is a healthy level of distrust to have with any conglomerate and their products, this includes Gen A.I. With that being said, if the technology exists, I could very well benefit from it in an ethical manner.

PEACE OUT!

Vaughn.