Project MayTH: Can AI manage a city?

The answer is yes, AI can go ahead, beyond managing a smart city.

I and my fellows have been researching & developing on a featured platform MayTH project since 2020. It’s not quite easy to talk about the completion of the project, of course. We have been also research and development a similar project for the transportation nodes, primarily airports. It’s called RABI (real airport business management) platform. Out motto is ” World’s First Unmanned Managed Airport concept”.

At MayTH project, What we have done so far?

Already collected 3,7+ TB data from the city pillars. Operations, consumptions, man-power and human resources etc. However we collect data from different cities (total 3), we created isolated data clusters and pools for the each city, then combined for the federated learning vectors.

It’s obvious that the most important point to develop a smart platform for a super system (such as a city) is creating almost seamless operational data models, and then collect the correct data and store all in other dynamic models, and lastly process it online/offline before the putting forward to the presentation.

We currently notice that (by observing the performance indicators) it’s possible to increase the efficiency up to %50 for an average city such as Istanbul city (approx. 5000sqkm, approx.20M pop.). It depends on several points that I mentioned above. But for the collecting the most appropriate (the best) data you have to have strong operational background. That means, knowing how the city is managed, operated and secured. Financial perspective, technical aspect, safety-security, citizen expectations and satisfactions etc. We have grouped the data subjects into 17 segments which are all inter-connected.

Digital Twin Side of MayTH

One of the other critical module of MayTH is digital twin platform (we call it as layer). But we mainly don’t bother to create (build) a digital twin of the zones from the stretch. System is featured to create inter-connected vectors to build the best (not ideal) twin model of the city zones. That’s fully dynamic model that many executions of the smart city operations are run over that MayTH-DT platform. Surprisingly, this module helps MayTH’s AI engine(s) to improve the “decision making under uncertainty” capabilities.

With the help of that module MayTH becomes more capable to execute many unmanned systems (vehicles, robots, interaction units etc.) real time. This is exactly the revolution of the Robotic era.

We could not find the chance to run all the MayTH features in a city yet, but DT learning models show us the efficiency in city operations (mostly touching the citizen life) can be improved up to %50. That’s not and cannot be actual, until we deploy the MayTH onto a city.

MayTH: AI as a Mayor?

Both Yes (why not as an assistant for today) and No (never reaches that ideal level). Japan’s tried

In my presentation to a huge global city management of a metropolitan municipality, I tried to clearly describe how AI (robots, assistants, whatever you would like to call it) can manage the city operations and manage the citizen life better than human does. It was around 30 minutes section of my presentation and I did not receive good feedbacks. Most gentlemen and ladies in the meeting were not happy with the potential future. Therefore, I realized they could not motivate to understand the technical, socio-cultural and socio-economical aspects of MayTH.

We clearly face the limits of the human’s weaknesses and strengths for the developing technologies. You can find forced positive articles published on Future of life platform . I can very well understand why these good-hearted and honest people aim to write and spread these articles. Really. But I also think they are obviously aware of that what is happening in other environments has already gone beyond what is written.

We have performed many algorithms (also tried particular decision makers such as Google Deepmind) so far to evaluate the better cognitive models so far. For example, in a trial run for a just 70TB size city-information (a 500K pop. European city) showed us very easy but hard-for-human-brain experience.

In processing only 9 months of operational data for this city, we encountered that almost 20% of the procedures actively uses are increasingly struggling to produce successful results. It’s not a indicator that human makes wrong decision, it showed us how our (an other) AI can do, overcome the very complex issues in a city.

It’s -surely- not limited to a city. But we know that, cities (all living zones) are a real super system comprising of many systems. Nevertheless, this AI modules and engines are possible to manage a factory, any kind of a facility such as airport, train station, university campus.

The Veins of MayTH: DMN (Mycelium)

I don’t want to put all the information again here. Digital Mycelium Network model is our new proposal to technology world. But firstly, it’s better to talk about the Mycelium and how it actually works! Then, can explain what is the primary model of DMN (Digital Mycelium Network) for MayTH.

Please don’t hesitate to contact me for any question or collaboration requests. We are open to the world.