My original written response:
“In this assignment, we were asked to choose a single idea from the readings provided and write a response. Since I have been in previous classes where the humans in the loop article were assigned, I focused mostly on reading the other article, “Power to the People: The Role of Humans in Interactive Machine Learning.” In the article, I wanted to respond to the topic brought up on page 110, “People Want to Demonstrate How Learners Should Behave”. I found this section of the article, in particular, to be interesting because it brings up a new dimension of interaction that is possible and that needs to be considered. We have plenty of examples in AI already of providing affirmation/correction to something an AI is doing, but there is more to explore in terms of types of interaction. I feel that there is an understandable satisfaction that can come from showing the AI what we want it to do instead of simple course correction.
Looking more specifically at showing it what to do, or giving it hints as to what to do instead of explicitly telling it what not to do.
Teaching by demonstration is a natural way that I have observed humans teach when dealing with not only AI but animals and even people. Teaching by demonstration is a natural method we as humans use in order to convey thought and show others what we know as it is done in practice which is not only easier to understand in some cases, but easier to demonstrate. In my experience as a human learner (student), I find that when I am learning something new, a demonstration can be the easiest way to get the idea across. I find this in the opposite position as well in my experience of sharing ideas, or teaching someone else about something, especially when it is a concept that is either highly involved or ‘second nature’, the teaching is extremely difficult to do without some kind of hand motion or demonstration of the topic in question.
When I was in cheer and dance, the choreography was almost always taught by repeated demonstration from the choreographer, occasionally giving corrections until the dance/movement was understood and being done ‘correctly’. I have found as a learner and a teacher is that when a process is so deeply ingrained in someone’s mind, be it choreography or even something simple like picking up a coffee mug, it can be incredibly difficult to teach beyond demonstration. Some actions or processes have come so naturally to a human being for so long that they have forgotten what it is like to not know, and find it exceptionally difficult to re-enter the mental space of a learner and are limited to demonstration until they can adjust into a mental space that allows them to communicate such a process in a more specific and relatable way.
That being said, we are more complex than the way we label the world and the judgments we make because of it. As much as we would like to, we cannot quantify our existence, and trying to have a human being teach in that format seems more likely to feel unnatural and frustrating. But by not only allowing a human to be in the loop of an AI system but allowing them to interact in a way that feels free and natural, could not only improve the practical use of this technology but the relationship we have with it as a whole. When we are designing these systems, let's not only put humans in the loop but humanity in it as well. “
ChatGPTversion of my response reworded for readability:
“In this assignment, we were tasked with choosing an idea from the readings and writing a response. Having previously studied the "Humans in the Loop" article in another class, I decided to focus on the other article, "Power to the People: The Role of Humans in Interactive Machine Learning." I found this article particularly interesting, especially the section on page 110 that discussed how people want to demonstrate how learners should behave.
This idea introduces a new dimension of interaction that needs to be explored and considered. While AI systems can provide yes/no answers or corrections, teaching them by demonstration is a more natural and satisfying way to convey knowledge. This is a method that humans have used to teach animals and even other humans. Demonstrating a concept in practice can often be easier to understand than simply explaining it.
As both a learner and a teacher, I have found that demonstrations can be the most effective way to convey complex ideas or processes. For example, when I was learning cheer and dance choreography, the choreographer would demonstrate the steps repeatedly until we understood and performed them correctly. However, when a process is deeply ingrained in someone's mind, it can be challenging to teach it beyond demonstration. It can be difficult to re-enter the mental space of a learner and communicate a process in a specific and relatable way.
While we cannot quantify our existence, we are more complex than the way we label the world and the judgments we make. Teaching in a format that tries to do so feels unnatural and frustrating. However, by not only including humans in the loop of AI systems but allowing them to interact naturally and freely, we can improve the practical use of technology and our relationship with it as a whole. When designing these systems, we should not only put humans in the loop but also humanity in it.” - ChatGPT of Me
For part two of this response, we were asked to come up with a list of 10 activities or tasks that could benefit from an interactive AI approach/mindset.