DAIQUAN

  • (made with electron, nodejs, html, scss, ml5, face++, azure, rekognition, and watson)

  • Concept + Domains

    What does it feel like when we are profiled and discriminated against by machines?

    DAIQUAN is an AI powered kiosk that takes the users personal information and tries to make their travel experience safer by screening for possible threats or harassment by local officials. It uses biased data for good.

    • Human-Computer Interaction
    • Biased Data
    • Machine Learning

    For this project I focused on some of the more popular API's that are currently available and used for facial recognition. These include Amazon's Rekognition, Face++, Microsoft's Azure, and IBM's Watson. The data that's extracted from each of these API's is then evaluated and compiled by DAIQUAN and used to rank the user.

    Form Focus





    The form was not hard for me to come up with. I knew from the beginning that I wanted to build something that related to transportation and data. Different places treat people differently based on the culture of that environment and how people are generally profiled. The idea for a kiosk came from the ticket machines that you see all over NYC in various train stations and bus ports.

    The sketches above show a wide range of some of the ideas I had initially gone through. This was when I was considering the input factor above everything else. The way the user interacts with the application was vital as it spoke to how the AI and the user interacted. The first form of the application required the user to submit social media information as part of the classification process but this raised some problems. Collecting the data from the accounts proved to be a bit problematic as it eleminated all those who didn't have a social media presence.




    Above is a simple mockup of what the initial interface looked like. I wanted the project to also take in bio data and use that as part of the evaluation. This idea came to me when I was traveling and saw the Clear Kiosks all over the secuirty portion of the airport. Here was a company that was using biometric information and data to process people and grant them/deny them access through the airport. My inital form factor was going to be a replicate of the Clear kiosk but I thought it best to change course and add my own flair to the project.

    Demo + System Flow




    The video above is an example of DAIQUAN in action. The video shows the application from the front-end perspective of the user and what they would experience. The user begins the process by entering their name and then pressing begin, after that DAIQUAN does the rest of the work. It snaps a photo of the user and then runs that image through the various API's that it's connected to. After that it builds a profile of the user and then asks them to confirm some information before moving forward. Once the user cofirms or denys the application then prepares the results of how they will be profiled and loads up ads that are targeted to the user. Based on just the visual information that it was able to collect.



    As seen in the chart above the application has just a few functions but each one works towards building the complete profile. The most important one is the function that builds the profile. It takes various things into account such as what the person is wearing, their gender, race, age, and even if they are wearing lipstick or have a beard. All of this is then compiled into a profile and then based on the final output it is compared with the crime statistics of NY giving the user a sense of how they'd be treated in NY. The profile is stored as a JSON file and DAIQUAN reads this file and generates a reciept for the user with all the inforation that was collected as well as wether or not they are flagged as a safety threat. The user can scan the barcode on the bottom of the reciept to learn more about the project.

    Branding for this Century

    So I'm building an application that uses AI and data to warn people about certain exposure and harrasment that they may recieve, how would this be branded? Will it have a serious tone or be more friendly? What generation is it geared towards? As I sat and thought about this on the train I came across a realization. A lot of the companies that I was seeing advertisements for had a similar approach as far as design and branding goes. It was as if they all shared the same agency. So I modeled the branding for DAIQUAN to look somewhat similar to theirs. Now that I had the general look down I started to think about the culture of these places and how some of them just may be trying too hard to fit in or be inclusive. You see it all the time online, companies are constantly being "cancelled" for their neglegent approach to advertising campaigns that often end up being offensive in one way ot another to a particular group of people.


    In order to replicate that pattern I programmed the loading/advertisement page to show specific content based on what the user was profiled as. These stereotypes come from my own personal experience as well as speaking with people from different backgrounds and ethnicities who told me how they often get misclassified and judged by people. So what DAIQUAN is trying to do is serve the user an ad that they might like based on the steroetypical data DAIQUAN has about this user.

    Physical Construction

    DAIQUAN is currently being built

    Next Steps

    DAIQUAN is currently being developed and will be on display at the MFA Design & Technology Thesis Show from May 14th - May 17th. Information about about the event can be found here.