Community Meeting (Jan 23, 2019): Ample Labs

Description

Presenters: CG Chen, Cheryl Li and Matt Wong

Ample Labs was born out of a pilot project called HomeTO and a desire to contribute to our local community here in Toronto. Their work was originally inspired by social initiatives like San Francisco’s Link-SF as well as broad research showing that even when experiencing homelessness, most people rely heavily on smartphones to find vital information.

Since then, the team has grown to include a dedicated and diverse group of designers, developers, researchers, consultants, public servants, and community members—all united by a common commitment to using tech for social good.

Website: https://www.amplelabs.co/

Notes

Video recording

  • Ample labs is a Social start-up
  • ChalmersBot
    • Web app to help people find local resources
  • Accelerator
  • 80~90% of people experiencing homelessness are invisible, in transition or not often counted
    • 90% own phone
    • 70% own smartphone
  • Started with user research
  • One researcher was inclusive design grad from OCADU
  • People tend to turn to Google, not services that are established. People don't know where to look or what's available to them
  • What is available is not easy to use
  • They created personas to capture their users more generally
  • Is a chatbot something people would want to use?
  • Is it easy to use?
  • Use the chatbot as a means to track what people are looking for
  • Initially started by partnering with Ada, but it was too rigid a tool for their needs
  • Amazon (Lex, Lambda) is another solution they're currently exploring
  • Getting data from
    • Homeless Help
    • 211 Toronto Data
    • Manual Entry
      • Corrections based on actual observations, etc.
    • Web scraper Tool
    • + on ChalmersBot
  • Demo of ChalmersBot
    • It gets your location to show you what's nearby for meals, shelter, etc.
    • Gives you directions to get there on foot (using Google Maps)
    • Asks for feedback and suggestions of things it should suggest or offer
  • Have seen about 500 unique users so far
  • Implemented analytics to track usage
  • Privacy concerns?
  • Question: Resources change, how do they address them other than checking manually?
    • They checked with the city for their process. The city checks twice a year to make sure everything they have listed is still valid
    • Maybe ChalmersBot can ask if the service accessed is still valid and the user can report themselves (crowdsourced)
    • How do they know when the city has updated? They don't know yet
  • Question: security and privacy? How do they handle the information? How is security being handled on either end?
    • Personal info: the only thing they can collect currently are their location and the thing they searched for
    • They have "eligibility" info that is collected like age, gender, sexual orientation, etc. and so they're working on that
    • They don't use the information
    • Security: not currently being worked on
    • Suggestions
      • Would be good to have a plain language terms and conditions to make it easier to use and understand
      • Hook ChalmersBot up through SMS?
        • Cost is restrictive
  • Question: how do you develop your personas? Do you see any limitations?
    • Distances you from the reality of people's lives, from individuals
    • They wanted personas to frame who users of the product would be in the beginning
    • Their lead researcher is suggesting to use mental models and move away from personas
    • Giving names to the personas could add more distance from the case, "historifies" it.
  • Questions from Ample
    • They work with Eloisa!
    • What are some best practices when it comes to Inclusive Design research?
    • How about design based on that research?
    • Does IDRC use personas?
      • We suggests using codesign instead, as a process
      • Personas have limitations
      • Use narratives, stories, real comments from real people instead
      • When you try to make a persona, you're summarizing, reducing a group of people that may not represent what they want in their life
      • Going through entire stories can take time, but it's okay, it's real, just publish all of them. If not everyone goes through all the stories, it's okay
      • In our practice, we would be losing richness and diversity of experience
    • Are there projects where there is a need for personas instead of individual narratives?
      • Might have been helpful in a case where it was a very technical project for requirements building. Even then, it's almost a dozen personas.
      • Personas can be a useful shorthand, but they abstract away specificity
        • Once you shear off the things that don't fit in a persona, you lose that extra information and requirements
      • Useful when you do have a clear set of functional requirements that need to be satisfied
      • Putting names and titles on them may not be the most useful
      • Instead of personas, maybe create vignettes or user stories and test the application within that specific situation
    • (missed a question about NLP and chatbot?)
      • Giving people the time to speak, if they speak slowly, etc
      • Going into a group, let them be the ones to make you feel comfortable rather than the other way around
      • Embedded codesign?
      • When Ample was going into the shelters, were there people from the shelters to facilitate the sessions?
        • Yes
      • Involve those groups right from the beginning. Build trust, they're part of your team, go back to them, get them to test your product, etc.
      • A codesigner's role will change from session to session
      • In certain locations, they don't want anyone from the outside to be there, so you would share materials and requirements with a person in the group and they will conduct the session themselves (e.g. teachers in a classroom)
    • When people have gone through trauma, how can you have them share that experience without forcing them to relive it?
      • Make sure you collaborate with people who are part of the organization
      • Getting emotional doesn't necessarily mean you've done anything wrong
      • You let that person lead, if they want to. You may not get everything you want, but you'll get something. Don't lead their discussion
      • Defer to psychologists, therapists, etc. who have more experience with this. If a person opens something up, could make sure they close it after, practice mindfulness, try not to leave people in the state they got to during the session
    • How can Ample learn more about codesign?
    • Certain services are only available if you qualify for them based on age, gender, sexual orientation, etc. How do you ask for that in a respectful way?
      • Don't use services that try to determine this information based on anything like name or anything else
      • Could just present the information you have with the eligibility criteria embedded and let the user choose themselves rather than restrict what's available
      • Good not to do it if you can avoid it
      • What the chatbot is doing makes this difficult to determine
      • Maybe before beginning the chatbot session, ask the user to fill out a pre-screening survey.
      • Make sure to include options like "prefer not to say"
      • Instead of asking up front, list the searching info and then allow the user to narrow it down based on various criteria
        • Or follow up with filtering questions
        • Maybe this filter can be stored
        • Maybe it doesn't need to be stored!
        • Ask before storing it
        • Store it in a cookie which is client-side only and not on a server
        • Not storing it on a server may alleviate privacy and data security concerns
        • Make it clear what is stored where
    • They meet in person at Civic Tech on Tuesday evenings and Sunday afternoons
    • Amplelabs.co, they have the people they're looking for at the bottom of their website
    • Would partnering with the IDRC be an option?
      • They met with Jutta to show her a beta version, still need to follow up
  • Accelerator
    • Program, not a product
    • Working with a local charity
    • Through their research, found that a lot of programs don't deal with homelessness from a prevention perspective
    • There's a lot of stigma around marginalized youth
  • Question: are there any plans for figuring out how organizations and charities could use this?
    • There's a feature called "Add a new resource", which is in beta
  • The tool is open source and on GitHub