The company where I am following my graduation internship is called GroupM. GroupM is a media agency specializing in digital advertising. They are the world’s leading media investment company. The location where I have been working is Eindhoven, within the Labs department.
Figure 1: GroupMs company vision
Figure 2: Company statistics year 2022
Figure 3: Organogram
Labs is an innovation department within GroupM in which they create co-creation environments where exceptional student talent explores disruptive trends and develop technologies that will shape tomorrow. Within Labs, I was placed in the team “Generative AI & Synthetic Media”.
The initial assignment for my graduation internship is to teach people about generative AI text prompts in order for them to effectively use it in their daily workflow. The main problem discovered during the project was that people have too little knowledge of generative AI and are in some cases frustrated and afraid to use this new technology.
Generative AI is an AI that mainly focuses on creating new content based on existing data. It starts with a simple description of what you want and will generate new content. This new content is called synthetic media. This technology has made big advancements over the years. The three primary factors that contributed to the growth are: improvement in models, better and more data, and greater computer capacity. (van Rijmenam, 2022) Generative AI has come to the point that it can produce high quality new content indistinguishable by humans. This AI can create various forms of content, including text, audio, video and more which makes up most advertising materials. This is why generative AI is interesting for GroupM, an advertising agency to use.
Generative AI can create new content faster than humans take to complete a task. It can be more efficient and is becoming better over the coming years. But generative AI at this moment is not perfect and employees of GroupM don’t know how to effectively use it.
Some of these frustrations and limitations are:
Although the employees of GroupM are very positive about using generative AI to make certain tasks more efficient but struggle with how to use this technology effectively and safely. The goal is to teach employees of GroupM about how to use generative AI effectively, in which work situations it can be used and what to watch out for when using this technology.
Figure 4: Generative AI content created with a text prompt using Leonardo.ai
The methodology used for this project is called Human Centered Design (HCD). This methodology prioritizes understanding and meeting the needs of the target audience, ensuring meaningful solutions. This methodology helped find the core problem employees at GroupM experience when using generative AI. HCD has three phases: Inspiration, ideation, and implementation. In each phase, I will describe how the project has progressed and the results of the deliverables.
Figure 5: Human Centered Design (HCD) methodology (Innovation Training, 2022)
The assignment was intentionally left broad. For this reason, the first three research questions were meant to define my project and who I was making it for:
| Questions | Link to document |
| What is generative AI? | Research 1: Generative AI & Synthetic Media |
| What impact/possibilities does generative AI have on the creative sector and marketing? | Research 2: Trend analysis |
| Which department(s) in GroupM Eindhoven is best fit for the implementation of generative AI in their workflow? | Research 3: GroupM workflow in Eindhoven |
With the results of each research I created my design challenge.
Based on this design challenge two new research questions were created
| Questions | Link to document |
| How do you create an effective training program that is fun and engaging? | Research 4: Training programs |
| What information do the employees of GroupM need to know/learn to effectively use generative AI in their daily workflow? | Research 5: Knowledge required for the training program |
Figure 6: Generative AI content created with a text prompt using Midjourney
Phase 1
In the inspiration phase, I gathered information about the context of the assignment and the target audiences, needs problems, and experiences with generative AI. With this information I created my project plan.
I held a debriefing with my company mentor which revealed that the assignment was intentionally left broad. This allowed me to choose the direction for the project in the defining phase. For this reason, I conducted research to explore the context. The first was a literature study about what generative AI (GAI) is. The second was a trend analysis to understand the relevance of this technology for GroupM. The main opportunity found is that GAI can optimize almost every phase of the creative process.
To define the target audience, I held interviews with five different creative departments within GroupM where GAI potentially could be implemented. The interviews focussed on what problems employees were experiencing in their workflow and their general opinion/knowledge of generative AI. I created an interview report and a new research in which I would visualize the general workflow of GroupM Eindhoven. Based on the results I concluded that the main problem was that employees within GroupM have insufficient knowledge to use generative AI effectively in their workflow. So instead of making workflows more efficient, I would focus on teaching the knowledge needed to use GAI effectively.
The target audience will be employees of GroupM in the creative departments that lack the knowledge to use generative AI effectively in their own workflow. To visualize the target audience personas and an empathy map that focussed on their pains & gains was created.
Figure 7: Persona 1
Figure 8: Persona 2
Phase 2
In the ideation phase first ideas and a concept to solve the problem was formulated. The final concept is defined and visualized in the concept book.
I began by brainstorming fun ways to teach the target audience the necessary information to effectively use GAI. I came to the conclusion of wanting to create a workshop and platform. The reason for choosing a workshop is that the target audience will get to practice with generative AI in a controlled environment and is easily accessible making it low entry for employees to join. At GroupM, training is often given through workshops. But generative AI is constantly evolving so for after the workshop there needed to come a platform that would easily scale with this technology and keep them continuously learning.
For the creation of the concept, I did research on how to create an effective training program through benchmark creation and attended multiple workshops within GroupM. I learned a lot from experiencing them myself. To acquire all the information necessary for the target audience I created a literature study, held a survey to test the target audience on their knowledge of text prompts, and held two days in the life to gather real-life scenarios for them to use GAI in their workflow.
For the eventual platform I brainstormed and came up with the idea of Weekly Challenges. To visualize this idea and the general flow of the platform I created a user flow with scenarios and created wireframes of pages. Tested the wireframes on its usability and validated the concept and experience with the target audience. Read more about the progress in Weekly challenges.
Figure 9: Brainstorm workshop and platform
Figure 10: Weekly challenges wireframe and high-fidelity prototype
Phase 3
For the implementation phase I focussed on the actual development of the workshop and weekly challenges platform.
Within GroupM there is a platform that showcases training sessions in which employees can enroll. These training sessions could be anything, I myself followed a Google Analytics workshop. This platform would be perfect to put the workshop online. I held an interview together with Anouk with one of the persons responsible for this platform and planned together with her a date when the workshop would be held. This would be the 25th of April.
As the date was set the workshop needed to be created. In my previous research I found multiple methods that would be beneficial for training people skills. For the workshop I used hands-on experience and a case study to let people practice writing prompts. The workshop would also address in which work scenarios GAI could possibly be used and the ethical & legal considerations. To create a successful workshop learning objectives were formulated and tested if they were achieved. Read more about the realization in workshop and validation in survey.
The platform's wireframes were tested and based on those findings small adjustments were applied in the high-fidelity prototype. This prototype was again tested with stakeholders and experts in the field. Based on these findings together with other conducted tests the prototype is found: modern, organized, user friendly and overall the target audience is very interested in this platform. A POC was created to eventually test the effectiveness of teaching the target audience through weekly challenges. However, with the limited time left, advice would be given instead on how to further test this prototype. Read more in Weekly Challenges.
For the eventual handover of the project I am currently working on a handover document in which I describe what has been created, what needs to be done, and advice for the future of this project.
Figure 11: First time giving the workshop together with Anouk Min
Figure 12: Realization of workshop and weekly challenges
