Skip to main content

AI basics: A Human in the Loop

 

AI generated text using a picture of kids in the library.

I am currently taking a Google AI basics course and have reached the "play with generative AI" phase. A central theme of the course is the importance of a "human in the loop" approach—emphasizing human oversight and the value of an iterative process.

Questions to ask when evaluating an AI workflow

  • Is the task generative? 
    • Yes. I was tasked with creating a social media post using generative AI tools—Gemini for the text and Gemini's image generation tools for the photo.
  • Can the task be iterated on to achieve the best outcome? 
    • Yes. For the photo, the tool asked for a selfie, but I uploaded a group photo instead. The AI flagged that it wasn't a selfie, but offered to crop/isolate one person from the group.
    • For the text, once I had the image I wanted, I asked Gemini for a caption prompt. The initial output didn't quite match the mood of the photo, so I described the image in more detail and refined the request. I then selected a concise paragraph from the generated options and fed it back into Gemini to generate relevant hashtags.
  • Are there resources to provide adequate human oversight?
    • Yes. Since I am modeling this for student teachers, I need to demonstrate effective oversight. As educators, they will act as the "human in the loop" for their students until students learn to independently critique AI output for accuracy, tone, message, and length.

The AI-Collaborated Post for the photo above

As we move further into the future, the ability to research, think critically, and self-direct learning will be the ultimate superpower. The habits built today between library stacks are building the critical thinkers, innovators, and leaders of tomorrow. #raisealearner, #librarymagic, #raisingreaders

The "Aha!" Moment

Teaching writing has taught us that fluency comes through practice—even if a student starts by writing "I don't know what to write" over and over. Generative AI operates on a similar principle: it depends on fluency of thinking.

While you can start an AI prompt with "I don't know..." or "Help me brainstorm...", you get far better quality in significantly less time when you begin with a human seed—an idea, an inkling, or a clear thought of your own.

Labels:

Comments

Post a Comment

Popular posts from this blog

Professional Practice Dissertation Pre-Proposal, Part 1 of 4

My dissertation proposal as a culture-based education (CBE) version of Chopped All-Stars Please indulge my need for metaphors and analogies to make sense of my world. This is Part 1 of  4 blog posts to clarify my thinking on my proposed dissertation topic.  How will this study work? Gather strong chefs, leaders and innovators in their own right and challenge them to create synergistic culinary masterpieces in the CBE Project,  a professional development program. change chef to teacher; change culinary masterpieces to culture-based education-infused practices and curriculum )  The parameters:  time (Kamehameha Hawaiʻi 4-week course with deadlines for teachers' own action research and learning portfolio to follow) key ingredients ( CBE practices , moenahā framework, makawalu ,  and the   National Writing Project program model) the course (teachers' own content area and current curriculum) The question: How doe...
6 comments
Read more

Dissertation Proposal 4 of 4: Timeline

Timeline Summer 12 - pre-proposal, lit reviews, submission for B-credit, negotiations for a course (place, time) Fall 12 - build the course using the village, build in  indigenous, developmental evaluation and sustainability Spring 13 - submit IRB approvals, pa ʻa the logic model, recruit teacher leaders Summer 13 - Alana Summer Institute Fall 13 - teacher action research and gathering of data Spring 14 - analyze data, write Summer 14 - write, collaborate, present This is a "in an ideal world" timeline that does not take into account the village, the permissions from higher ups, the ability to exist on a non-budget, the recruitment of teachers willing to participate in this work with me, the full time job duties that I already have AND the fact that Iʻm supposed to be in 6 or 9 credits worth of my own courses in the Summer of 13 on Oahu when I want to serve my Big Island teachers. Any suggestions?
3 comments
Read more

Dissertation 3 of 4: Research Methods

"From the vantage point of the colonized, a position from which I write, and choose to privilege, the term 'research' is inextricably linked to European imperialism and colonialism. The word itself, 'research,' is probably one of the dirtiest words in the indigenous world's vocabulary." ( Tuhiwai Smith , 1999) Paula Moya, a Chicana researcher says, "identities are fundamental to the process of all knowledge production" ( p. 102 ).   Like Tuhiwai Smith, I must also embrace my identity as a colonized researcher.  Moya goes on to say that education,  "should not be about merely inculcating status quo values,"  but to reject the status quo as a way to inculcate a "transformative multicultural education" that will educate all learners for democracy and social justice  (p. 109). The Alana Project, starting with this professional development course, which becomes my dissertation action research, sees transformative multicu...
3 comments
Read more