How Uber’s Predictive Machine Learning Is Changing User Experience?

Uber has continually invested in AI over the years, and the technology is now widely used as a critical component throughout the company. “Uber has been an AI powered company for years and years and years,” said CEO Dara Khosrowshahi. Uber has made substantial use of artificial intelligence (AI) and machine learning (ML) to maintain its leadership in the transportation business.

Uber operates in over 10,000 cities across more than 70 countries, handling 25 million trips daily with 137 million monthly active users. Machine learning (ML) is integral to almost every interaction within Uber’s apps, enhancing various aspects of the user experience and operational efficiency.

Phase 1 (2016-2019): Predictive Machine Learning

During this initial phase, Uber primarily focused on predictive machine learning for tabular data use cases. The company employed algorithms like XGBoost for critical tasks such as ETA predictions, risk assessment, and pricing. Uber also ventured into deep learning for 3D mapping and perception in self-driving cars.

In 2016, Uber acquired Geometric Intelligence to create Uber AI Labs, demonstrating its commitment to advancing AI research. Gary Marcus, founder of Geometric Intelligence, stated:

“We live in an era where data is central to almost every major company and for almost any of them, even incremental improvements in artificial intelligence and machine learning can have immediate, significant payoffs,”

Uber began its machine learning (ML) journey in 2015, with applied scientists developing models in Jupyter Notebooks™ and engineers designing customised pipelines for production deployment. Initially, there was no standardised approach for developing reliable and reproducible pipelines for large-scale training and prediction operations, nor was there a simple mechanism to store and compare training experiment findings. Putting a model into production necessitated the creation of a specialised serving container.

Uber developed Michelangelo in early 2016 to standardise machine learning procedures, allowing developers to construct and deploy ML models on a large scale. Michelangelo addressed the issues of scalable model training and deployment, and later launched Palette, a feature repository that allows teams to organise and share feature pipelines in batch or near-real time. Palette now includes over 20,000 characteristics that Uber teams may use to develop effective ML models.

Key Michelangelo components include:

Gallery: A model and ML metadata registry with a comprehensive search API.

Manifold: A model-agnostic visual debugging tool for ML.

PyML: A framework to expedite prototyping, validating, and productionizing Python ML models.

Horovod: A tool for distributed training.

Phase 2 (2019-2023): Deep Learning and Collaborative Model Development

Between 2019 and 2023, Uber dramatically improved its AI and ML capabilities by adopting deep learning (DL) and encouraging collaborative model creation for high-impact ML initiatives. During this revolutionary period, Uber shifted from considering model iteration as code within a centralised ML monorepo to supporting DL as a first-class citizen in Michelangelo, the company’s centralised ML platform. Michelangelo received significant improvements, including broad support for deep learning models, infrastructure for training, deploying, and monitoring DL models, and a unified user interface (UI) and code-driven experience. This flexibility benefited both traditional ML and deep learning developers, allowing for more efficient model construction.

Uber also established Project Canvas, which applies software engineering approaches to machine learning development. This guaranteed that all source code, configurations, and dependencies were kept in the ML monorepo, which allowed for change tracking, code reviews, and unit testing. The project provided both local and remote training, allowing developers to debug and test programmes locally before submitting them to remote clusters for complete training. By the end of this phase, more than 60% of Uber’s tier-1 models had included deep learning in production, considerably improving model performance across a range of applications. These included ETA calculation, rider-driver matching, fraud detection, and personalised recommendations, all of which saw significant accuracy and efficiency gains using deep learning models.

Additionally, Uber’s Customer Obsession Ticket Assistant (COTA) used machine learning and natural language processing (NLP) to improve customer service. COTA, which was built on Michelangelo, helped address more than 90% of inbound support queries in a timely and effective manner. When a new ticket was created in the customer service platform, the appropriate features were gathered and delivered to Michelangelo’s ML model. The model predicted scores for each feasible solution, which were later kept and accessed by the backend service. Agents were presented with the top three alternatives for ticket resolution. This method cuts ticket response time by more than 10% while maintaining or increasing customer satisfaction.

Zoubin Ghahramani, Uber’s Former Chief Scientist from 2016 to 2020, emphasized the importance of AI:

“Artificial intelligence and machine learning are absolutely central to Uber’s mission. Uber’s opportunities are unique among major technology companies, because they center around the real physical world, which is complex and difficult to predict. We have to navigate around the real world, develop perception and action systems for our self-driving cars, and understand, predict, and make more efficient the experience for our riders and drivers. At a larger scale, we are trying to model and optimize entire cities, and reimagine the future of transportation through, for example, urban VTOL aviation.”

Phase 3 (2023-present): Generative AI

Uber’s current phase of expanding its AI/ML capabilities indicates a significant move towards incorporating Generative AI to improve user experiences and internal operations starting in 2023. The Gen AI Gateway is at the heart of this endeavour, providing a uniform interface for accessing both external large language models (LLMs) and Uber-hosted proprietary models. This arrangement enables Uber to combine the vast knowledge and complex reasoning of external models with the targeted accuracy and performance of in-house solutions for specific jobs.

The Gen AI Gateway’s key capabilities include complete logging and auditing to track LLM usage, cost management features with over-consumption alarms, adherence to safety and policy rules, and PII redaction to secure personal data prior to contact with external LLMs.

Uber’s generative AI applications span a variety of domains:

Internal Productivity: Tools like Copilot automate repetitive work for developers, produce code snippets, and aid with mistake correction and unit testing, allowing them to focus on more complicated development chores. Generative AI also helps create and evaluate design papers more effectively.

End-User Experience: AI-powered chatbots improve customer service by resolving frequent requests with human-like interactions, resulting in higher satisfaction and operational efficiency. The DragonCrawl system uses LLMs for mobile testing, responding to UI changes dynamically.

Business Operations: AI models improve Uber’s fraud detection skills by analysing trends and identifying fraudulent activity, hence increasing platform security. AI-powered synthetic data systems increase the quality and reliability of linear learning models used in product development.

Michelangelo’s Generative AI Extension: Uber has expanded its Michelangelo platform to enable LLMOps (LLM Operations), which includes responsibilities such as fine-tuning data preparation and quick engineering, as well as LLM deployment, production environment service, and continuing performance monitoring.

Personalising the UX is crucial to keep the app relevant for each user and to minimise navigation effort. At Uber, we utilize machine learning in crucial parts of the booking process, such as product selection. The machine learning system identifies the appropriate list of products to display to a user based on their usage history, current market conditions, and personal preferences. For instance, if a rider frequently uses Uber Auto, the system will prioritise showing Uber Auto in various app sections., said Madan Thangavelu, Senior Director, Engineering at Uber.

Throughout these phases, Uber has developed and implemented various AI/ML initiatives. Recent advancements include extending AI/ML infrastructure with CPU- and GPU-centric solutions, such as Nvidia’s H100 GPUs for generative AI applications, which aim to improve latency and throughput. Uber’s deliberate move to cloud infrastructure, which began in February 2023, improves the scalability and efficiency of AI programmes such as Generative AI.

Educational efforts like the ML Education Programme demonstrate Uber’s commitment to provide its employees with the skills and knowledge needed to effectively adopt and scale ML solutions, guaranteeing sustained innovation and competition in the transportation business.

Uber has integrated a comprehensive AI and machine learning strategy throughout its operations, benefiting several elements of its company. Marketplace Optimisation is a crucial endeavour that uses machine learning techniques such as LSTM networks to anticipate supply and demand patterns in real time. These projections allow Uber to proactively send drivers to high-demand locations, increasing trip counts and revenues.

“We kind of look at every life, every phase of the general lifecycle and how we optimize that with AI.” –  Uday Kiran Medisetty, Distinguished Engineer at Uber. 

Dynamic Pricing, powered by machine learning algorithms, modifies rates in real time based on customer demand and driver availability. This surge pricing strategy helps to balance supply and demand during peak hours, resulting in effective service delivery while maximising income prospects.

Dispatch optimization algorithms employ machine learning to properly match passengers with drivers, taking into account parameters like distance, traffic conditions, and user preferences. These algorithms anticipate millions of matches each minute, increasing efficiency and cutting wait times for both passengers and drivers.

Fraud detection and prevention is another major area where Uber uses machine learning models to detect and prevent fraudulent activity. By analysing transaction data for abnormalities and suspicious trends, Uber improves security and protects both drivers and riders from fraudulent transactions.

Uber’s DeepETA project uses deep learning to optimise routes and predict ETAs. This project increases the accuracy of ETA predictions by combining historical and real-time data, assuring optimal route planning, and improving the entire user experience.

Computer vision plays an important role in improving map accuracy and navigation. Uber employs machine vision to update maps in real time, which improves destination prediction and routing accuracy. Deep learning models also provide 3D mapping and vision in self-driving technologies, allowing autonomous cars to navigate safely and efficiently.

These AI and machine learning efforts demonstrate Uber’s dedication to using sophisticated technologies to optimise operations, improve customer experiences, and drive innovation throughout its worldwide platform.

Despite the closure of Uber AI Labs in 2020 due to pandemic-related cost-cutting measures, Uber continues to invest heavily in AI and ML technologies. The company’s current focus is on integrating generative AI into various aspects of its operations, from developer productivity to customer experience.

Lior Ron, founder and CEO of Uber Freight, recently emphasized the transformative potential of AI in logistics:

“Very soon, we will see AI completely revolutionizing logistics. Once things are digitized and once things are connected, then magic happens on top.”

From 2023 until the present, Uber has made a concerted push towards Generative AI, employing sophisticated LLMs to improve internal efficiency and end-user experiences. The creation of the Gen AI Gateway and the extension of Michelangelo to support LLMOps capabilities have enabled Uber to continue developing and preserving its competitive edge in the transportation business. As AI technology advances, Uber is committed to investigating its potential to alter urban mobility and beyond.

Dara Khosrowshahi, CEO of Uber, addressed the company’s plans with Generative AI:

“I think that the earliest and most significant effect that AI is going to have on our company is actually going to be as it relates to our developer productivity. Some of the tools that we’re seeing in terms of Copilot are going to allow our devs to kind of be super devs and to be able to innovate more, build more, faster.”