Leading Change in Crisis Management: Our DIMA Chatbot Success Story in Collaboration with DataCamp
May 9, 2024
Aerial view of the Hatay Province in Türkiye, on February 19, 2023. [State Department photo by Ron Przysucha/ Public Domain from Flickr]
Introduction
In response to the escalating severity and frequency of global natural disasters, our innovative collaboration with DataCamp through the AI Innovation Challenge has led to the creation of DIMA (Disaster Information and Management Assistant), an AI-driven chatbot specifically designed for disaster zones.
The solution described in this success story is a testament to how technology, when applied thoughtfully, can transcend traditional disaster management approaches, offering real-time, actionable information, and even crucial but responsible emotional support to those in urgent need. This narrative explores DIMA’s journey from concept to deployment, highlighting its impact and the sophisticated technology behind its success.
The Critical Need for Tailored AI Solutions Beyond General AI like Chat GPT
In today’s rapidly changing environment, having access to timely and precise information is vital, not only during emergencies but in daily life as well. General AI chat systems like ChatGPT offer broad guidance on various topics, such as preparing for natural disasters or what essentials to pack in an emergency kit. However, these systems typically operate on fixed datasets that might not always provide the most current details or information specific to a user’s real-time needs.
Recognizing this gap, we developed our custom chatbot designed for both general and crisis-related queries. Unlike standard AI solutions, our chatbot leverages tailored, up-to-the-minute data to deliver crucial information pertinent to any situation—be it locating nearby services or navigating emergency scenarios. This capability to offer precise, targeted assistance enhances individual preparedness and peace of mind across various contexts, significantly mitigating potential stress and confusion. By integrating our carefully curated data into the chatbot, we extend beyond the limitations of conventional AI, crafting a tool that’s invaluable for everyday challenges and indispensable in critical situations.
The Problem
When disaster strikes, every second counts
Often, phone lines are jammed and help can’t get through quickly enough. That’s where a special kind of helper, a custom chatbot, comes into play. Imagine a computer program designed just for emergency situations, ready to answer questions, give updates, and guide people to safety 24/7. This article will show you why these chatbots are not just helpful—they’re essential. They can talk to thousands of people at once, keeping everyone informed and calm, which really can make the difference between chaos and order. Let’s dive into how these smart chatbots work and why they are a must-have in disaster response efforts.
Imagine that an emergency happens, phones are down. What happens next?
The Goal
Our mission was to develop a custom AI-powered chatbot within an ambitious eight-week timeline, specifically tailored to streamline and enhance disaster response for a designated city or area.
This project transcends traditional recovery efforts—it’s about building resilience and instilling hope in communities shattered by catastrophes.
By harnessing advanced Artificial Intelligence, we are not merely adapting to challenges; we are preempting them, creating a system that is uniquely equipped to handle the specific needs and nuances of the affected area. Our initiative underscores the necessity for bespoke AI solutions that elevate the effectiveness and efficiency of disaster response, ensuring that help arrives faster and more accurately when it matters most.
The Background
Disasters of the past years that would have needed such a solution that inspired this project
The catastrophic earthquakes in Turkey and Syria in February 2023, which registered magnitudes of 7.0 and 6.5 respectively, underscored a critical vulnerability in our global disaster response capabilities. The devastation was profound, with thousands of lives lost and countless others displaced, pushing emergency services to their limits and revealing a dire need for enhanced support systems.
Compounded by the growing threats posed by climate change, which is expected to increase the frequency and severity of natural disasters worldwide, the urgency for advanced disaster management solutions has never been greater.
The Partner
Partnership with DataCamp: Empowering Innovative Solutions Thanks to Education
Our journey to create a groundbreaking disaster response chatbot was significantly enhanced by our partnership with DataCamp. Through their generous DataCamp Donates program, we accessed a treasure trove of educational resources that brought our vision to life. DataCamp provided our team with an entire year of professional training, covering a wide array of skills directly applicable to our project. From learning about cloud-based technologies to understanding data-driven decision-making, DataCamp’s courses were pivotal. This partnership didn’t just supply tools; it empowered our team with cutting-edge knowledge and skills, setting our project on a path to success.
Our Approach to This Problem
0. (Additional Preparations for the participants)
Provided by Datacamp
Our partnership with DataCamp was essential. They provided us with access to educational courses that improved our team’s skills in areas that were crucial for building a responsible and trustworthy solution.
1. Primary Preparations
Understanding the Main Purposes:
- Emergency Response Assistance: Modeled after PSAP operators and 9-1-1 dispatchers, these chatbots are designed for clear, concise, and directive communication. They are engineered to handle urgent, action-oriented requests efficiently, mirroring the operational demands of emergency situations.
- Emotional Support Assistance: Inspired by the roles of psychologists during crises, these chatbots focus on providing empathetic, supportive, and open-ended interactions. They offer comfort and understanding to those experiencing psychological distress during emergencies.
What should it know?
User Questions Cathegorisation
To decide on our chatbot’s capabilities, we categorized potential user questions into seven key areas, each with a brief explanation and an example:
- Knowledge. Offers information about earthquakes, “What causes earthquakes?“
- First Aid. Provides procedures and guidance for emergency medical situations, “How do I treat a broken arm until help arrives?“
- Safety and preparation. Share tips for staying safe before, during, and after earthquakes, “How should I prepare my home for an earthquake?“
- Information and contacts. Details on the locations of hospitals, shelters, and Red Cross centers, “Where is the nearest relief shelter located?“
- Alerts. Gives timely updates on earthquake activity, “Is there an earthquake warning for today?“
- Location. Offers functionality to give users instructions on how to reach safety, “How do I get to the nearest hospital from my current location?“
- Emotional Support. Offers advice for managing stress and fear related to earthquakes, “How can I manage anxiety after an earthquake?” This feature aims to support mental health by providing tips for emotional coping and resilience.
These categories helped refine the chatbot’s scope by pinpointing specific functionalities and information it should offer within the project’s timeframe. Moreover, they directed our development efforts towards serving targeted user groups more effectively. The potential questions also played a key role in identifying the data needed to train and develop the chatbot.
Decision to use Psychological Profiles in order to create a responsible mental-health related solution
- Communication Style Alignment: Following Kern’s research, the chatbots’ communication strategies are aligned with the psychological profiles typical of their human counterparts in similar roles. This alignment ensures the interactions are not only personalized but also highly effective in meeting the specific needs of users during emergencies.
- Research-Driven Approach: The design of our chatbots is supported by findings from Kern’s study, which uses social media data to predict personality traits and values. This research backs our method of matching chatbot communication styles with the psychological characteristics of different occupations.
2. Developing the Chatbot
Step 1: Collecting the Right Data
Firstly, we gathered a wide variety of crucial data that our chatbot might need during a disaster. This includes information like safety protocols, emergency contact numbers, and maps showing resources like shelters or hospitals. Collecting this data is vital because it helps the chatbot give accurate and useful advice to people who need help.
Step 2: Organizing and Checking the Data
Once we had all the necessary data, we put it through a process to organize and check it. This involves:
- Sorting the Data: We made sure all the data was named and arranged in a consistent way, which helps in managing it effectively.
- Removing Duplicates: We cleaned out any repeated information to keep our database tidy and efficient.
- Validating Information: We checked that the data was correct and up-to-date, such as making sure maps had the right location names and coordinates.
Step 3: Making the Data Understandable and Efficient for an AI System
After organizing the data, we used some smart techniques to prepare it for use:
- Text Processing: We set up systems to categorize and refine the text data, ensuring that the chatbot can understand and use it properly.
- Data Integration: We used advanced tools to help the chatbot process different kinds of data more effectively, which improves how it understands and responds to queries.
Step 4: Integrating Real-Time Alert Systems
To ensure timely and personalized alerts, we integrated a real-time notification system that evolves around:
- Push Notifications: Depending on user preferences, the system sends alerts through email, SMS, or in-app notifications, ensuring users receive critical information immediately.
- Location-Based Alerts: Utilizes user-provided zip codes or geolocation data to send relevant local updates, facilitated by a cron job that monitors for new data around the clock.
Step 4: Building the Chatbot’s Response System
Our chatbot uses what we call a “retrieval-based” method to decide how to answer questions. It looks through all the data it has, finds the most relevant information, and then forms a response based on what it finds. This method helps ensure that the chatbot’s answers are both accurate and helpful.
Step 5: Taking Care of Context-Aware Personalized Communication Styles
We customized the chatbot to communicate in different ways as mentioned before to make it as trustworthy and as responsible support solution as possible.
To describe how we did it technically we used language processing (NLP) techniques. By leveraging machine learning models trained on large datasets of conversational texts, DIMA was at that stage able to read subtle cues in user language and tone. This allows it to categorize personality types effectively.
For instance, it can act like a 911 dispatcher, giving direct emergency instructions, or like a psychologist, offering support and guidance to those in distress.
Look at the screenshots below of the “behind the scenes” prompt.
You can see similarities and maybe even some pro tips for your own situations while using chatbots or simply ChatGPT.
For the psychologist persona, the prompts are structured based on the field operation guide for training first-aid disaster psychologists, published by the National Center for PTSD and National Child Traumatic Stress Network. This approach emphasizes equipping individuals with the skills to overcome obstacles themselves, promoting sustainable long-term recovery from disasters.
Step 6: Going Beyond Generic Solutions
We chose to build our database and not rely solely on generic AI models like GPT for a reason.
Our specialized chatbot not only answers general questions but also provides specific, up-to-date information tailored to each user’s needs. This bespoke approach is particularly important in emergencies, where timely and accurate information can save lives.
Through these steps, we’ve not only built a chatbot but created a life-saving tool that stands ready to assist when disaster strikes. This project is a testament to our commitment to innovation, precision, and most importantly, safety. Our chatbot isn’t just a technological achievement—it’s a beacon of hope and reliability for those facing the chaos of natural disasters.
Step 7: Integrating Mapping Features
Understanding that users may need to find safety quickly in an emergency, we integrated mapping features into our chatbot. This tool helps users find the nearest emergency services, like shelters or hospitals, and guides them there step-by-step:
- Accessing Real-Time Maps: We connected our chatbot to mapping services, which use satellite and street data to show current maps on users’ devices. These maps help users see where they are and where they need to go.
- Providing Directions: Once the chatbot knows a user’s location (either through automatic GPS detection or manual entry of a ZIP code), it calculates the best route to the nearest safe location. It then gives step-by-step directions, much like a car navigation system, guiding users whether they are walking or driving.
- Interactive Map Features: Users can interact directly with the map through the chatbot. For example, they can zoom in/out and view different routes, helping them understand their journey better and make informed decisions quickly.
These mapping features make the chatbot not just a source of information but also a practical guide, especially useful in times of crisis when finding help quickly can be crucial.
Step 8: Continuous Improvement and Updates
We constantly update and improve the chatbot’s systems and databases to keep its responses relevant and timely. This includes using the latest tools for managing and retrieving data, which helps the chatbot respond quickly and effectively during ongoing disasters.
Benefits and Other Applications these Methodologies
Applied Methodology 1:
Chatbot’s feature to act based on personality types
Can also be utilised in:
Customer Service
- Benefit: Enhanced interaction quality by adapting communication styles to match customer preferences increases satisfaction and loyalty.
- Application: AI tools analyze customer personality types from initial interactions to tailor responses, ensuring each communication is personalized and effective.
Human Resources
- Benefit: Improved hiring accuracy and team dynamics by matching personalities to job roles and existing team profiles.
- Application: AI analyses from candidate interviews are used to recommend matches based on the personality fit for specific teams or roles, optimizing workforce composition.
Marketing and Sales
- Benefit: Increased effectiveness of marketing campaigns through messages that resonate with various personality types.
- Application: Marketers use AI to segment audience data by personality traits, enabling highly targeted and personalized marketing messages that drive engagement and conversions.
Healthcare
- Benefit: Better patient compliance and satisfaction through personalized communication.
- Application: Health professionals use AI to adjust their communication methods according to patient personality types, making advice more understandable and actions more persuasive.
Finance
- Benefit: Enhanced client trust and satisfaction by providing personalized financial advice.
- Application: Financial advisors use AI to analyze personality-derived risk tolerance and investment preferences, tailoring financial advice to individual client needs
Education
- Benefit: Improved student engagement and performance by tailoring learning experiences to individual needs.
- Application: Educational platforms employ AI to modify teaching methods and materials according to the personality types of students, optimizing learning outcomes.
E-commerce
- Benefit: Increased customer satisfaction and sales through personalized shopping experiences.
- Application: E-commerce platforms integrate AI to analyze shopper personalities and tailor product recommendations and displays accordingly, enhancing the shopping experience.
Mental Health Services
- Benefit: More effective therapeutic interactions by understanding client personality types.
- Application: Therapists use AI tools to preliminarily assess client personalities, which helps customize therapeutic approaches and improve the effectiveness of treatments.
Applied Methodology 2:
A Chatbot that can give very precise, real-time updates or changing quickly information
Can also be utilised in:
Real Estate
- Benefit: Offers instant updates on property status, new listings, and market conditions.
- Application: Real estate agencies use chatbots to inform clients about new properties on the market, changes in property availability, and urgent updates in real estate trends.
Event Management
- Benefit: Real-time coordination and information dissemination during events.
- Application: Event organizers implement chatbots to provide attendees with updates on schedules, venue changes, or cancellations, enhancing the event experience and reducing staff workload.
Tourism and Hospitality
- Benefit: Updates tourists on travel information, enhancing their experience.
- Application: Hotels and tourist centers use chatbots to inform guests about local events, weather conditions, transportation schedules, and any last-minute changes.
Transportation and Logistics
- Benefit: Streamlines updates on shipment statuses, traffic conditions, and delivery schedules.
- Application: Logistic companies implement chatbots to provide instant updates on the status of deliveries, traffic reports, route changes, and estimated arrival times.
Sports and Entertainment
- Benefit: Provides real-time updates on game scores, player stats, and event news.
- Application: Sports leagues and entertainment venues use chatbots to keep fans informed about game developments, behind-the-scenes content, and event-specific announcements.
Time Frame
In just 8 weeks, our team achieved something remarkable: we built a highly advanced AI-powered chatbot from the ground up.
This outcome not only shows our commitment to using the latest technology to solve real problems but also highlights our ability to move fast and deliver results when it really counts.
Completing such an ambitious project in under two months is a testament to our team’s skill, speed, and dedication to making a difference when it matters most. This achievement showcases our prowess in tech innovation and sets a new benchmark for rapid development in the industry.
The Outcome
- Comprehensive Information Delivery: Beyond basic data, the chatbot offers critical information such as the locations of relief shelters and safety updates in real-time.
- Integrated Mapping Feature: Seamlessly integrates with our application’s mapping tools, guiding users to the nearest shelters with driving instructions, which is crucial during evacuations.
- Personalized Alert System: Features a robust alert system that sends timely notifications tailored to users’ preferences, which can be received via email, SMS, or in-app notifications. This ensures users receive relevant information based on their location or selected zip code.
- Real-Time Monitoring: Utilizes a cron job for continuous monitoring, ensuring that the chatbot provides the most current and pertinent information to users during ongoing disasters.
- Scalable Infrastructure: Capable of handling a high volume of interactions simultaneously, which proves essential during widespread crises.
- Data-Driven Insights: Employs a comprehensive and regularly updated database to ensure all communications are both accurate and relevant.
Further Possibilities
- Multilingual Support: Expanding the chatbot’s language capabilities to better serve non-English speaking communities and enhance its utility in global scenarios.
- Predictive Analytics: Integrating advanced predictive models to foresee user needs and provide preemptive information and suggestions, potentially before users even request it.
- Enhanced Emotional Intelligence: Increasing the chatbot’s ability to recognize and respond to emotional cues for more empathetic and supportive interactions.
- Blockchain Integration: Implementing blockchain technology to secure data exchanges and ensure user data privacy, enhancing trust.
- Integration with IoT Devices: Connecting the chatbot with Internet of Things (IoT) devices to improve data collection and response accuracy in real-time.
- Advanced Geolocation Features: Enhancing the chatbot’s geolocation capabilities to provide even more precise location-based services and directions during emergencies.
- Continuous Learning: Employing machine learning algorithms to enable the chatbot to learn continuously from interactions, thereby increasing its accuracy and effectiveness over time.
Author of an article: Weronika Dorocka – VP of Business Development
Want to work with us too?
