Teacher Advisor

Leveraging IBM Watson to help teachers save time while lesson-planning.
 

BACKGROUND

Teacher Advisor is a free resource that helps K-5 teachers plan lessons. Using IBM Watson, Teacher Advisor sifts through professionally vetted content, allowing teachers to find lessons, activities, and teaching strategies catered to their own classroom needs.

OBJECTIVE

Our team was tasked with finding opportunities for further development of Teacher Advisor so that it directly targets the challenges that K-5 math teachers face while lesson planning. After conducting research and analyzing our findings, we defined our problem statement as:

How might we help teachers save time while lesson planning and find materials that are customized to their students’ specific needs?


My contributions: 

I conducted desktop research, interviewed 3 out of our 8 participants, affinity diagrammed the research findings with my team, and led the development of our persona, Ted the Teacher. I also illustrated the storyboard of Ted’s day (keep an eye out for the sticky note saga gif below). Here's a video of me presenting this storyboard and persona to these stakeholders, including Phil Gilbert, the head of design at IBM.

The features that we developed as an outcome of our process are now being implemented within Teacher Advisor.

What I gained:

Counter to my social sciences background, I realized through this project that— as informative as raw data can be— the most relatable way for humans to make sense of data is through a story.

I also gained an understanding of the value of zooming out on a project and considering the system that surrounds a user. For example, while teachers are the primary users of Teacher Advisor, we learned through our research that teachers’ administrators, students, and families are also integral players in this problem space.


That’s me in the middle. The shirt was a bit loud, in hindsight…

That’s me in the middle. The shirt was a bit loud, in hindsight…

PROJECT DURATION

August 2017 - December 2017

This project was developed under a jointly taught internship-course by the University of Texas at Austin & IBM Design. More information about this course can be found here and here

THE TEAM

Christy Zhang, Gene Azad, Me, Anna Brink, Sneha Jain, & Mauricio Herrera


Generative Research

commoncore.png

SECONDARY RESEARCH

From our desktop research, we gained a better understanding of the current landscape of lesson planning, which involved familiarizing ourselves with the Common Core, what teaching standards are, and how they differ among states. Not only did this information inform our primary research objectives, but it was also crucial to understanding how Teacher Advisor currently organizes content.

 
Our prioritization grid for research objectives.

Our prioritization grid for research objectives.

PRIMARY RESEARCH

We then used a prioritization grid to establish as a team where the most critical gaps were in our understanding of how teachers lesson plan. Based on our high risk, high uncertainty quadrant, we decided that the most important questions to investigate through further primary research were:

  1. What do teachers consider a successful lesson plan, and what do they consider a failure?

  2. Where are teachers, physically, when they lesson plan?

  3. How do teachers feel at each step of the lesson planning process?


Interviews

Objectives

  • Understand what teachers consider a successful lesson plan

  • Understand which tools/resources teachers currently use to lesson plan

  • Understand the approximate timeline of the lesson planning process

Method Justification

  • Allowed us to ask users followup questions about how they think and feel about the process and existing resources

  • Best alternative to contextual interview (which we could not do due to time and geographic constraints)

Method Details

  • 8 participants total

    • 5 - elementary school teachers

    • 1 - K-5 science teacher

    • 1 - 6th grade math teacher

    • 1 - child of a teacher

  • 2 in-person interviews, 6 phone interviews

  • Asked participants to send a picture of their lesson-planning workspaces to supplement our understanding of their context

The desk where one of the teachers we interviewed completed her lesson planning.

The desk where one of the teachers we interviewed completed her lesson planning.

Analysis & Findings

  • (analysis and overall findings shown below under' ‘Sensemaking’ section)

  • Lesson Planning timeline:

planning-timeline

Competitive Analysis

Objectives

  • Understand how existing tools/resources help teachers lesson plan

Method Justification

  • Allows us to understand key strengths and weaknesses

  • Quick, low-cost method

Method Details

  • Used responses from interviews from teachers as a starting point

  • Included direct, partial, parallel, and analogous competitors

Findings

Of the most popular, digital tools/resources that teachers used, the most common strengths were:

  • Ability to create an account/profile to save content

  • Feedback from other teachers

  • Ability to organize content

Most common online resources that teachers mentioned using.

Most common online resources that teachers mentioned using.


SENSEMAKING

To synthesize the data that we collected throughout our research, we used affinity mapping to organize trends, identify themes, and uncover insights. We then came up with 3 summary insight points:

 

1) Teachers often feel like they have too many resources for lesson planning, which are overwhelming and time-consuming to sift through.

2) The most time-consuming part of lesson planning is making modifications for each individual student's needs.

3) More often than not, teachers take their work home with them.

My teammates and our almost-finished affinity diagram.

My teammates and our almost-finished affinity diagram.

 

User Persona

To better empathize with our users and their pain points, we decided to create a user persona.

Our user persona, based on our research findings.

Our user persona, based on our research findings.

A glimpse at Ted's weekly schedule, based on our research findings.

A glimpse at Ted's weekly schedule, based on our research findings.

Storyboarding

A day in the life of Ted the Teacher:

ted-as-is
 

Here’s a video of me presenting Ted the Teacher's As-Is Scenario to a live audience.

 

Pain Points

From our research, insights, and journey mapping, we boiled Ted's pain points down to 2 key issues: 

 

time

The time pain point was distilled from the interviews with teachers and the interview with a child of 2 teachers. They expressed that it is very difficult for teachers to maintain their desired work-life balance because they have little time available during their work day for lesson-planning. “Planning periods” are often filled with mandatory departmental or administrative meetings, and as a result, actual planning must take place after work, displacing time spent with family members. 

“We would move our schedule around for when he’s grading, so if you needed help with your own math homework or something, you better make sure you asked before he gives a test, otherwise he’s gonna be busy.”

— child of 2 teachers

customization

Public school classrooms are typically comprised of students with many different needs: some students require more attention in math, some students may need specific fonts for dyslexia, while others may require calmer activities. Teachers must adapt their lesson plans for each of these students’ individual needs, and these adaptions are called 'modifications' in the teaching world. While teachers may be able to reuse lesson plans from previous years, these modifications must be made for each lesson and require additional planning time.

“How I figured out what works for each kid? Well I would talk to other teachers, try one thing, see if it works for that kid, and if not, try another thing… pretty much just trial and error.”

— K-5 science teacher

ted-needs

To-Be Scenario

From these two pain points came two overarching goals for our concept:

Ted can find material customized for his particular class makeup using Teacher Advisor, in less than 10 minutes.

Ted is able to complete lesson planning at school so that he can maintain and develop his family relationships.

teds-to-be

Prototyping

With these two goals in mind, we created a low-fidelity paper prototype and used Wizard of Oz methods for some quick user testing:

teacher-advisor-paper-prototype.gif
user-testing.jpg

After several iterations, the solution that we developed was a customization tracker in the Teacher Advisor planning interface. These small, color-coded icons would allow Ted the Teacher to keep track of whether specific modifications had been met. This solution also allows Watson to perform searches tailored to each teacher's classroom-specific needs and predict supplementary material that Ted the Teacher may need for his upcoming classes.

 

Here is a breakdown of how the customization tracker works:

 
 

Customization Tracker

Upon logging in for the first time, users are prompted to build an initial classroom profile based on individual students’ needs.

Some student needs might include: ‘English Language Learner,’ ‘Above Level,’ ‘Below Level’ (terms gathered from research)

 
 

Default Auto-Fill

Watson generates a lesson plan that best fits the classroom profile. Each activity is tagged with potential modification requirements that it may fulfill, based on previous teachers’ use of Teacher Advisor.

This one’s not a gif ;)

This one’s not a gif ;)

 
 

Watson Suggests

If users want to remove an activity and replace it with another, they can find more similar activities that fit their classroom profile needs by clicking on the Watson icon.

watson-suggests.gif
 
 

Lesson Feedback

Upon logging in the next time, users are prompted to provide a quick, yea-or-nay review of each activity to allow Watson to more accurately predict users’ future lesson-planning needs.

feedback.gif
 

Future Directions

If we had the opportunity to further develop this concept, the next step we would have taken would have been creating a higher-fidelity prototype and testing it with several teachers in situ, ideally over a whole school semester. 

One potential issue that we foresaw with this concept is the rate at which students' needs change, versus the rate at which Watson can learn a teacher's needs. If students' needs change quicker than the rate at which Watson can learn, then our concept would need some heavy revision. This does pose a rather interesting, socio-technical question though: is there a way we could teach Watson to learn teachers needs, quicker than students typically change in their learning needs?