Week 2: Community Driven AI

Week 2: Community-Driven AI

 This week I devoted spare time to an AI Lesson plan strictly based on Open AI’s criteria for scholars. I then explored Microsoft’s resources to advance my grasp on AI Strategy. 

If data is the new oil then one could assume that Artificial Intelligence (AI) is the new electricity (source). The problem is it needs a lot of data and capable talent to support getting it going in any organization. Many organizations are betting on AI strategies in the future, but here in the present, we have many cases of automated decision systems used in the public sector that is failing us. My question is always, “who is monitoring and evaluating these systems ensuring they are doing the job?” Just like you get your car check one would expect getting algorithms used that impact people evaluated would be the norm. Sadly in the public space, this is not true. This had me thinking again. Who better than the community and organizations we have in place to do this? I believe community-driven AI will accelerate innovation in the same way open-source has done for the internet. I think it’s essential to building reliable systems and should include not only engineers but different specialists in the decision making the process. This week I looked at New York who was the first state to pass Local Law 49 which created an Automated Decision System Task Force to do this. This is what I consider a light review of what can go wrong or right in community-driven AI.

Hype vs Innovation

Based on the chart above we are at a point in the hype cycle and technology adoption cycle where AI accessibility and availability will become as commonplace as software as a service subscription model. Typically when you want to know where innovation is happening you look at where the money is going. In venture capital funding for AI is continuing to increase and at the same time, we are seeing research happen at the academic level and be deployed in the private sector quite quickly. When profit is an incentivize innovation happens but we should be wary of that as well. A look at several organizations with non-profit missions that are well funded show a growing group of folk who also care about this. For example, if we look at groups like Open AI or even AI Now who released a paper on government use cases of automated decision systems we are finding orgs that understand the impact of the technology.

There are also many failed projects. One project that is ongoing that has faced critique by many is the NYC Task Force. Established by Local Law 49 to organize a task force to study how government AI impacted, people. The Task Force brought in an interdisciplinary team of scholars, lawyers, computer scientists and more yet struggled to define the work they do. They are expected in Fall 2019 to launch a paper but many critiques have pointed out that they have been slow to do any impactful work. This is a big lesson for the urban innovation community because it reflects the first state effort to get something right. While I don’t know the details or why it failed I can only speculate that working with government is hard and organizing a loose set of leaders without cash-incentivize to do work is particularly difficult.

(AI Business School)

While all hype isn’t the same there are breakthroughs in AI happening. Over the last 3 years Microsofts, for example, has done some significant work here. Microsoft even has a repo and well-written documentation to get people or organization started in using the framework to train then deploy solutions. One of my favorite data visualization tools I see cities using is PowerBI. Still, what will this innovation look like if governments and communities take part in shaping and keeping it accountable? I think the open data portals of cities are exciting because it gives chances for regular people whom we could call Civic Hackers to take the data to create tools to help their cities. In that same way, I think we can use data and these tools which can create new forms of engagement data to build better AI solution.

How can we get AI into communities? I think it’s about literacy and understanding of how to deploy the technology. Often times for decision makers that means identifying what problem they are having that this can solve. Most times they won’t have enough data to train a model. In those scenarios, I usually like to look at data that is sufficient. For example, helping organizations pair public data sources like social media, chatbots or newsletter data is one of the simplest ways I think to deploy AI. Most people can use these things to support customer services or even sales initiatives which can help lower cost and provide value. As community-driven AI becomes less of an idea and more of a reality I expect more states or even local government to organize around this. I see communities becoming more in tune with the risks and flaws of AI that even big firms struggle with. Throughout my time I’ll explore more cases like these and continue to leverage Microsoft AI repository to work on proof of concepts.

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 50% of AI Business School from Microsoft.
  2. I explored Open AI Scholars work and scholarship work plan.
  3. I played around with Microsoft AI Repository and quick deploy documentation.
  4. I researched cities who have AI organizations.

References: