Week 3: How NLP Can Help Local Government

Week 3: How NLP Can Help Local Government

This week I got to sit in on some meaningful conversation with large organizations in the city. We have some exciting things brewing here at Microsoft Miami team. This week  I explored Natural Language Process (NLP) and it’s use and potential impact in Urban cities. I reference much of Stanford CS course and follow the works of AI Researcher, Fatma Tarclaci.

Communication has ushered in perhaps the most transformative forms of use cases in Artifical Intelligence (AI). Language is at the center of this and that’s why I think Natural Language Process (NLP) which is a subclass of AI will continue to innovate even at the local level across cities. I am no AI expert I studied Business for that reason I think being proficient in strategies to implement cutting edge technologies is important. The problem is AI talent is expensive and the underlying infrastructure it takes to implement it within an organization isn’t accessible especially for cities. Fortunately, at Microsoft, we have a platform and team of the worlds best working on this. We even released a free AI Strategy through AI School.  In my work this week, I looked at Natural Language Processing which is a combination of computational linguistics and computer science understanding how this was impacting local communities for good and bad.

How NLP has been used for decades

“Natural Language Process in laymen terms to me is taking computer systems who process human language and turn it into meaning.”

If you ever used Microsoft Word – like the billions of folk across the globe you probably know what the little red/green line under a word means. This is a form of natural language processing called text correction. If you think about the impact that proper grammar and spelling have had on your life I want you to think about what advanced forms of this technology can do for people across cities. Through the internet several billion dollar companies who control online media have access to vast forms of media including text. Most of these are accessible through APIs with tutorials that easily let you use NLP for an application like sentiment analysis. One startup ZenCity used it to help city managers gauge how their cities are responding to using social media and other data channels. Employers could potentially find your handle name and run a simple analysis across all your social media to see the type of things you post about. The applications for a business run wild and policies around how this data is used is needed to keep accountability.

Microsoft Bot that leverages it conversational AI, for example, has guidelines that state:

  1. Articulate the purpose of your bot and take special care if your bot will support consequential use cases.
  2. Be transparent about the fact that you use bots as part of your product or service.
  3. Ensure a seamless hand-off to a human where the human-bot exchange leads to interactions that exceed the bot’s competence.
  4. Design your bot so that it respects relevant cultural norms and guards against misuse
  5. Ensure your bot is reliable.
  6. Ensure your bot treats people fairly.
  7. Ensure your bot respects user privacy.
  8. Ensure your bot handles data securely.
  9. Ensure your bot is accessible.
  10. Accept responsibility.

The guidelines above highlight how companies can task themselves even without policy to be responsible in the technology they use. Communication is key to instructing developers and business stakeholders. Chatbots are an interesting form of NLP application because of how ubiquitous, Apple’s Siri, but how much of us rely on these to get our work done. These have invaded our homes and our cities in more than one way. Even without these smart devices if you look at your phone and look at your apps like Facebook the microphone feature may be added where services listen to what you say then finding targeted ads by brands that fit the category or term you said. It’s real and it can be scary which is why guidelines are needed.

How Cities Play A Role

In my last week of writing, I explored how community-driven AI was a possible method for pushing innovation on accountability on AI usage at the state or local level. As cities become smarter well that is the perfect opportunity for evaluating these systems and automated decision systems. I see applications of NLP being essential to moving that forward. Communication can drive better experiences for people by helping them find affordable housing, navigate health benefits, or understand how to pay for their traffic ticket. As long as it deals with human interactions NLP can help.

“Those in position to profit are incentivized to accelerate the development and applications of systems without taking the time to build diverse teams, create safety guardrails, or test for disparate impacts. This is why we are arguing for greater funding for public litigation, labor organizing, and community participation as more AI and algorithmic systems shift the balance of power across many institutions and workplaces. -AI Now Report 2018, pg. 42-43

Cities have loads of data and automated decision systems that are outdated. In a report by, AI Now 2018 they dived into some of the dangers of these systems. The conclusion focused on many things voicing the need for an accountable system and interdisciplinary training of professionals. Natural Language Processing because of its nature in the humanities of language is a perfect introduction of how you can tie teams with diverse backgrounds to evaluate systems being set in place. For example, Uber recently apologized online on Twitter after sending a tweet that offended people. If we explore that we learn the danger wasn’t in any advanced form of NLP but rather a simple auto-responder bot.

The AI Now report hinted at bringing power back to the public because of the dangers of the private sector which focus on profit. I picked one company public blunder too touch on the fact that while Uber may get millions of tweets at them using a bot to auto-responder is the best way to keep customers happy. In the case that went wrong due to lack of oversight, we see offensive and customer outrage (or even lack of faith) in the company. Those less technical might have thought a human posted that leaving them to think Uber hires racists for example. In another case, we see how Amazon fires folk for not being productive through an automated decision system an algorithm.

A call for a diverse and interdisciplinary team to work on projects related to this, not just computer scientist is also needed. That’s because at the center of our sectors with unique problems we need people of all backgrounds to support bringing solutions. It’s not easy work, but someone has to think about it. Across sectors use cases of AI go far and wide in the depth of ways you can use algorithms to help workers be better or worse. In my study of NLP, I see a lot of hope in this to continue helping people and organizations.

I hope to leverage my learnings across the community to advocate for ideas that can scale locally and across our cities. With Microsoft, we have the potential to achieve scale across cities. In my work, I seek to look at urban innovation through the lens of tackling problems faced by innovation communities who lack resources to deploy artificial intelligence in internal operations. I’m looking for organizations and people who are seriously interested in this. Interested? You can reach me on Twitter or we can set time to have quick coffee if you’re in the greater Miami area.

Written by Gregory Johnson

Things explored this for personal learning this week:

  1. I went through 75% of AI Business School from Microsoft.
  2. I explored AI Nows 2017 and 2018 report.
  3. I played around with ChatFuel and Microsoft Chatbot framework.
  4. I began work on a framework and examples of AI use in the public here in Miami spreadsheet.

References: