Simple Explanation (ELI5)

Imagine having one digital ticket for buses, trains, bikes and scooters, like an assistant picks the best option for you. Artificial intelligence organizes schedules and routes and personalizes real-time directions. In short: mobility becomes a smart service, not just a purchase.

Key Advantages & Current Limits

🟢 Pros: AI improves planning, demand forecasting and real-time personalized solutions. 🟢 Pros: MaaS can integrate multiple modes into one service and reduce car ownership. 🔴 Cons: Most tests happen in wealthy countries and clear rules, inclusivity and data protection are needed. 🔴 Cons: Algorithmic bias and lack of interoperability standards emerge.

Future Outlook & Takeaways

In the next 3-5 years we will see more integrated MaaS and AI in cities, with real-time planning and personalized services. But for emerging markets, open standards, clear rules and inclusive data will be essential. Digital twins and ethical frameworks for greener, fairer mobility will likely emerge.

MaaS and AI: How Artificial Intelligence Changes Transport?

Article

100%

Imagine opening an app and seeing every option to move around the city — buses, trains, bike sharing, scooters — integrated into a single digital ticket. That's the promise of Mobility as a Service (MaaS): not owning a car, but buying mobility as a service. A new systematic review, published in *Future Transportation*, analyzed over 100 studies from 2020 to 2024 to see where we stand.

Researchers split AI integration into four levels. The basic level uses machine learning and classic optimization for schedules and routes. Moving up, blockchain and big data enter for security and massive data handling. Deep learning, however, is everywhere: it spans all levels but shines at the advanced stage for real-time personalized solutions, like "leave now, take bus 5, arrive in 20 minutes".

There's a catch: most trials happen in wealthy countries. Potential for emerging markets is huge, but we need clear rules, inclusivity for those without smartphones or fast internet, and ironclad protection of sensitive data. The road toward equitable, smart, and green mobility is mapped out: now engineers, lawmakers, and citizens must walk it together.

Bonus

100%

The review follows PRISMA guidelines on Scopus and Web of Science databases. The four-level taxonomy (Basic, Intermediate, Advanced, Full Integration) maps architectural complexity: from simple algorithmic journey planning (Level 1) to real-time multi-modal orchestration with smart contracts on blockchain (Level 4).

At the basic level, regression algorithms, clustering (k-means, DBSCAN), and linear/integer programming dominate for fleet optimization and demand forecasting. The intermediate level introduces hybrid recommender systems (collaborative + content-based) and reinforcement learning for dynamic pricing. The advanced stage sees the explosion of deep learning: LSTM/Transformers for short-term demand prediction, GNNs (Graph Neural Networks) for complex transport network modeling, and federated learning for training models on decentralized sensitive data (GDPR-compliance). Blockchain (Hyperledger Fabric, Ethereum) ensures transaction audit trails, identity management, and revenue sharing among MaaS operators without a trusted third party.

Critical gaps emerged: algorithmic bias on historical data (e.g., low-income areas underserved), lack of interoperability standards (open APIs, GTFS-Realtime, NeTEx), and adversarial robustness (adversarial attacks on routing models). Future work: human-in-the-loop MaaS, city digital twins for policy simulation, and ethical AI frameworks for inclusive mobility.