Blog

Guest Authors: Utpal Mangla, VP & Senior Partner; Global Leader: IBM’s Telecom Media Entertainment Industry Center of Competency at IBM, & Luca Marchi, AI Innovation, Center of Competence for Telco, Media and Entertainment, IBM, & Kush Varshney, Distinguished Research Staff Member, Manager at IBM Thomas J. Watson Research Center, & Shikhar Kwatra, Data&AI Architect, AI/ML Operationalization Leader at IBM

AI Fairness

Artificial Intelligence (AI) is becoming a key cog in how the world works and how it lives. But the reality is that AI is not as widespread in critical enterprise workflows as it could be because it is not perceived to be safe, reliable, fair, and trustworthy. With increasing regulation, concern about brand reputation, burgeoning complexity, and a renewed focus on social justice, companies are not ready and willing to deploy a “science experiment” at scale in their operations. As Thomas J. Watson, Sr., an early chief executive of IBM said, “The toughest thing about the power of trust is that it’s very difficult to build.”

We’ve seen many newsworthy examples of AI producing unfair outcomes: blacks being discriminated against in criminal recidivism, low-income students systematically having low “predicted” exam scores when the coronavirus pandemic cancelled the real exam, men and women having different lending decisions despite having exactly the same assets, and many more. Why is this happening and what can we do about it?

Lessons from Commercial Aviation

It is instructive to look at the history of commercial aviation to understand what is happening with AI today. The first flights by the Wright brothers and Santos-Dumont during 1903-1906 to the introduction of the commercial jetliner, the Boeing 707, in 1957 can be considered as the first 50 years of aviation. This period was all about just understanding how to make planes fly with limited commercialization. In the second 50 years of aviation that followed, the fundamental nature of airplanes did not change—today’s commercial jets are basically the same as the Boeing 707—but there was a heavy emphasis on safety, efficiency, and automation. Now commercial airlines operate almost everywhere with safety records hundreds of times better than fifty years ago.

What is the lesson for AI? We are just at the beginning of the second 50 years of AI. We can trace the beginnings of AI to a 1956 conference at Dartmouth. We can say that the first 50 years concluded when deep learning won the ImageNet competition in 2012. Just like in aviation, the first 50 years were spent on getting AI to simply work—to be competent and accurate at narrow tasks—with limited commercialization. Now our job is to work on making AI more safe, reliable, fair, trustworthy, efficient, and automated, and bring commercialization everywhere.

Accuracy Isn’t All You Need

To make AI trustworthy, we need it to be more than accurate. We need it to be fair so that it doesn’t discriminate against certain groups and individuals based on their race, gender, or other protected social attributes. We need it to be reliable and robust so that it can be used in different settings and contexts without spectacularly falling apart. We need it to be explainable or interpretable so that people can understand how AI makes its predictions. We need it to realize when it is unsure.

The LF AI & Data Foundation’s three open source toolkits: AI Fairness 360, AI Explainability 360, and Adversarial Robustness 360 Toolbox, are means for practicing data scientists to address these needs for making AI more trustworthy. Let’s dig into fairness in more detail. Where do the problems come from and how can we mitigate them?

Where Do Fairness Issues Come From?

AI, specifically machine learning, tends to reflect back and sometimes amplify unwanted biases that are already present in society. There are four main reasons why there can be unfairness in AI:

1. Problem misspecification – when the problem owner and data scientist pose the problem they are going to be creating a solution for, they may make choices that introduce unwanted behaviors. For example, if they want to predict whether someone will commit a crime in the future, but they design an AI system to predict whether someone will be arrested in the future, they can introduce unfairness. First, being arrested does not imply that a person is guilty of a crime. Second, there are more arrests made in neighborhoods where police patrol more often, and that is not done equally.
2. Features containing social biases – some attributes in a dataset already contain traces of structural biases that provide systematic disadvantage to certain groups. For example, the SAT score may be used as a feature for predicting an applicant’s success in college, but it is known to already contain biases so that some minority groups do worse because of cultural knowledge embedded in the questions.
3. Sampling biases – sometimes datasets overrepresent privileged groups and underrepresent unprivileged groups. For example, face attribute classification datasets are known to be skewed towards white males.
4. Data preparation – one key step in AI development pipelines is feature engineering, where raw data is transformed before being fed to the AI. There are several subjective choices made in this process, some of which can lead to unfairness

Measuring and Mitigating Unfairness

Just as there are many reasons why AI can yield unfairness, there are many ways to measure it and mitigate it. Choosing how to measure unfairness is not as easy as it sounds because different fairness metrics encode different worldviews and politics. As one option, you can measure the difference in selection rates of an AI, say the difference between the fraction of black applicants who got accepted to a college and the fraction of white applicants. As a different option, you can measure the difference in accuracy rates between the same two groups. They both sound about the same at face value but are actually quite different. In the first option, you are implicitly assuming that features have social biases (like the SAT score), but in the second option, you assume that all the unfairness is due to other reasons like sampling biases.

If you measure that an AI system is behaving unfairly, what can you do about it? You can apply one of many possible bias mitigation algorithms. The basic idea of bias mitigation is that you want a sort of statistical independence between protected attributes like ethnicity or gender and the predicted outcome like success in college. Statistical independence is the notion that two dimensions are unlinked and have no relationship with each other. There are many statistical methods that encourage independence, but that is a longer discussion for another day. Feel free to check out the AI Fairness 360 documentation for more details about bias mitigation if you can’t wait!

LF AI & Data Resources

• Learn about membership opportunities
• Explore the interactive landscape
• Check out our technical projects
• Join us at upcoming events
• Read the latest announcements on the blog
• Subscribe to the mailing lists
• Follow us on Twitter or LinkedIn
• Access other resources on LF AI & Data’s GitHub or Wiki

Guest Author: Susan Malaika, LF AI & Data Trusted AI Committee Member

On April 28, 2021, the LF AI & Data Trusted AI Principles Working Group hosted a session about applying the eight RREPEATS principles to two examples; one in the context of network providers, the other in banking. The Practical Examples session is a follow-up from the session in February that introduced the RREPEATS principles. 

After the presentation of the two examples, a round table discussion took place, covering the application of trusted AI principles in companies. A key outcome from the discussion was the recognition that low-level tactical tools can be used to implement trusted AI principles. Another outcome was the importance of having an ethics board in corporations.

Example: Classification of Encrypted Traffic Application

Iman Akbari Azirani and Noura Limam from the University of Waterloo in Canada, and Bertrand Mathieu from Orange Labs in France, explained that the Internet data payloads and, increasingly, headers are encrypted. Typically, 85% of Internet traffic is encrypted, and Google services are 100% encrypted, making it difficult to classify network traffic. In order to anticipate workloads and offer good services to customers, network operators need to understand the traffic by:

• Providing accurate capacity planning
• Detecting fraud, such as, attempting to mask data that would normally be paid for, as free traffic. A common example of fraud is transmitting video over a text service.

Bertrand, Noura, and Iman discussed three of the RREPEATS principles in the context of the encrypted traffic application; Equitability, Privacy, and Explainability.

To ensure equitability and privacy, significant amounts of information (e.g., temporal information, IP addresses, payloads, security certificates etc.) are removed from the data. For explainability, it is important to understand that classification is applied at the global level (service and application) and not at an individual level. Deep neural networks are used in classification and focus on three types of features (3 faceted models): Transport Layer Security (TLS) handshake bytes, traffic shape (size, direction, inter arrival times), and statistical features. In spite of being focused on protocol agnostic features, the classification accuracy of this approach is high.

This section ended with request for tools that:

• Help with network classification 
• Support efficient/speedier models running in the network
• Offer better explanations 
• Prevent attacks on model 

Example: RosaeNLG Framework (an LF AI & Data project)

Ludan Stoecklé, CTO of Data & AI Lab BNP Paribas CIB and initial author of RosaeNLG, outlined how to explain a decision to a non-expert user, focused on trusting the decision and being transparent. Apparently, it has been shown that non-expert users most often find textual explanations easier to understand rather than the output from corporate style dashboards, such as tables and graphs. Also, computer generated texts are often preferred over human written text, as the generated texts are clearer, less ambiguous, and more concise.

Ludan illustrated his points via credit application rejection and showed simple text generated via Natural Language Generation to explain the reason for rejection. The following considerations apply in preparing an explanation for a decision:

• Interpret the decision
• Define what to say
• Define how to say it 

The RosaeNLG framework (an LF AI & Data Sandbox project) provides template-based Natural Language Generation. It automates the production of relatively repetitive texts based on structured input data and textual templates, and is widely used in the financial industry.

Dataset Discussion 

There was an important discussion about open datasets in the areas of network encryption and NLG, and the absence of open networking datasets that could move the field forward. Typically networking datasets are proprietary and complex. It is difficult to create a synthetic dataset that mimics real network traces, with many sessions from many users and with variable network conditions that end-users experience. However, Orange is investigating the possibility of opening a part of a completely anonymized dataset (and without any possibility to reverse engineer it and discover private information).

Round Table – Applying Trusted AI Principles in a Corporation

The first part of the round table discussion was focused on the examples presented. An observation that principles have different meaning in the context of different examples and that understanding the context of different scenarios is key. Souad Ouali, the Trusted AI Principles Working Group Chair, responded that the RREPEATS principles were very carefully and deliberately worded to be applicable in a general way in international and varied settings. The round table panelists agreed that Trusted AI principles must be incorporated into the DNA of what we do throughout the AI life cycle and it is important to ask questions relating to principles at every step. An important observation from listening to the two practical example presentations is how tactical low-level tools map to high-level Trusted AI principles.

The round table ended with a discussion on corporate ethics boards and whether they should be geographically diverse or distributed, incorporate domain experts, and extend into the operational aspect of a business. Another consideration is whether an ethics board should include representatives from other institutions.

Join Us for Our Next Session

Please join us for The Trusted AI Principles – Tools and Techniques webinar on September 15, 2021. Register here!

Stay connected with the Trusted AI Committee by joining the mailing list here and attend one of our upcoming meetings! Learn more here.

RREPEATS Links

• April 2021 Trusted AI Principles – Practical Examples Webinar Session recording / Session slides
• RREPEATS – An Introduction to the Principles for Trusted AI – Thoughts and Next Steps
• LF AI & Data Announces Principles for Trusted AI

LF AI & Data Resources

• Learn about membership opportunities
• Explore the interactive landscape
• Check out our technical projects
• Join us at upcoming events
• Read the latest announcements on the blog
• Subscribe to the mailing lists
• Follow us on Twitter or LinkedIn
• Access other resources on LF AI & Data’s GitHub or Wiki

Guest Author: Dr. Jagreet Kaur, Chief AI Officer, Xenonstack

What is Data Intelligence?

The world is leading towards data-driven intelligence. To stand in the world of evolving technology and the competition phase, organizations must make data and AI-based decisions. It becomes difficult for the organizations that are not working on those aspects and data to know the facts and insights while making decisions.

Data Intelligence enables the process of multisource data and generates meaningful insights that would help to make valuable decisions. It allows combining unstructured data and text analytics results with structured data to use for predictive analytics. It can give a real-time statistical analysis of structured or unstructured data to understand data patterns and dependencies.

Why Do You Need It?

Data intelligence is required to process and understand the data. Data intelligence is rapidly becoming one of the most important elements of big data. Data intelligence has progressed from the infantile stage to a point where it can handle vast amounts of data with intelligence. It isn’t going to fold its wings either; the immediate positive results have attracted many organizations’ attention. Various entrepreneurs have expressed interest in using and developing data intelligence to make intelligent decisions in driving their business. There are multiple cases where we may need it. Some instances have discussed that helps to know why we require data intelligence:

• Artificial Intelligence: Using a machine learning algorithm helps to find the predictive analysis and to recognize correlation. It helps to find domain-specific custom entities and word usage.
• Intuitive Visualization: It allows us to understand data effectively in less time using informative intuitive charts and graphs. Visualization helps to understand complex data within seconds rather than reading and understanding an excel or any other data file. Visualization also generates insights and clear data patterns that are difficult to find in tables or datasets. It enables to easily filter and drill down the reports according to the requirements.
• Insight generation: Based on the collected data, it allows generating or taking insight from the visualization that helps to understand the business progress and customer needs.
• Data-driven decision-making: To make better and data-driven decisions so that those correct decisions can be taken to gain customer satisfaction and revenue.

The Base Foundation For Data Intelligence

Data intelligence is an optimized way that provides an unconventional 360 view of the business environment. It helps to understand the customer requirements better and also monitors the organization’s performance. Based on the data or insights, make decisions according to customer preferences and improve its revenue and benefits. Data intelligence is based on several sets of techniques accordingly to enrich business decisions and processes. These are:

• Descriptive (“What happened”): It is used to review and examine the historical and real-time data to understand business performance and customer behavior. It detects a particular occurrence of a situation.
• Diagnostic (“How it happened”): To know the reason for the occurrence of a particular instance or situation.
• Predictive (“What could happen”): It uses historical and, based on that, predicts future occurrence using some ML algorithm.
• Prescriptive (“What Should We Do”): To develop and analyze the alternative knowledge that can be applied in the course of action. It helps us to understand what to do in the future.
• Decisive: Decisive analytics helps to measure data suitability. It chooses the recommended action to implement it in the environment and real-time process when there are multiple possibilities.

How Can I Use Data Intelligence?

Data intelligence performs the following steps to identify relations and mentions of unstructured or structured data.

• Data Ingestion: It collects structured and unstructured data from different sources such as documents, emails, databases, websites, and data repositories. Data can be inserted into the application or platform manually or scheduled at fixed intervals of time. Data can be processed and used by that application to perform tasks.
• Data Processing: Now, data collected from sources can be processed and used to generate insights. It makes it possible to find a relation between data. Several tools provide an easy-to-use interface for creating custom models to train and test the model to find entities and data relations. It allows using models for future predictive analytics.
• Reporting and Visualizations: Reporting and Visualization is the final step that analyzes the data using charts and graphs. Visualization makes it easy to understand large and complex data effectively.

The Benefits Of Data Intelligence

Data intelligence gives wings to the technology by providing intelligence in their daily tasks and decisions. Let’s discuss the benefits of data intelligence and why organizations should embrace them:

• Changing demands: Data intelligence makes the organization adopt the dynamic changes of the industries. The business nowadays is continuously evolving. To stand in the competition and reduce the chances of failure, organizations must accept and update the newly emerging trends. For example, the trend of the adoption of selfie cameras in smartphones was increasing. Mobile businesses that do not capitalize on the trend are doomed to fail. Data intelligence helps organizations to understand customer behavior and change. Firms are informed about repeated changes and the pattern of occurrence by smart adaptive dynamics. It allows the company to make informed decisions based on the analysis.
• Strong Foundation of Data: Data intelligence makes big data more strong and strengthened by restructuring the process of data arrangements. It allows to gather insights from big data and then render optimized engagement capability.
• Useful Data: No doubt the world is generating a large volume of data every day that can change the world and improve the services according to customer demands and preference. But most of the data is not in the form of use. It is not possible to directly use that data. It is required to transform it into a useful form to use data. Data intelligence is also in charge of converting raw data into cumulative information. Data intelligence cleans and transforms data into smart capsules of ready-to-use data that can be used in the company to assess results. Data intelligence makes it possible not to worry about defining particular cases to the computers.
• Augmented Analytics: Advanced statistical approaches are used in data intelligence to advance visualized predictive and prescriptive analytics. Instead of building a complete application every time, it automates the data processing that can be completed just by doing some simple steps. If required, further changes may be recommended based on the results. There is no way for business plans to fail with such extensive planning for a real-life scenario. Advanced simulations enable businesses to predict potential outcomes and make changes to prescriptions as needed.
• Accelerate innovation: Data intelligence makes it possible to accelerate innovations by making smart use of data. It allows using data insights to drive business innovations and use them to develop their services by considering customer preferences and requirements.

What is the difference between data information and intelligence? 

• Data: It is the raw form of data recorded truth at a point in time. It might be a conversation, a purchase, or an interaction with your company’s website. Data is the compilation of results from those incidents that are then quantifiably recorded so that companies can review them easily. 
• Information: It is a collection of data or a way of bringing data together. When data is picked from an event and put into narrative forms, it helps to answer the following questions: 
  • What is the churn rate of employees?
  • How long is the sales cycle of an organization?
    The information helps to answer these questions that move the business.
• Intelligence: Intelligence is a group of information to derive intelligence or decisions in their application or tasks. For instance, suppose you are selling more in southern regions, then the smart and intelligent answer will be why that might be. To get an answer, it will look at numbers such as the number of events, amount spent on advertisements, marketing campaigns southern region clients receive. After that, it can be compared with the other region (North region). Through this analysis, we get to know that there are more client interactions in the southern region, so to increase sales in the North region, it is necessary to do the same.

Data Intelligence In The Real World

• Healthcare: Rapid digitalization of healthcare systems are adopting technologies to create a connected healthcare environment. Hospitals need to synchronize with the technology to become smart, advance, and more accurate. Hospitals use various types of sensors, apps, digital equipment that are generating a large volume of data regularly. This data can be used to automate several processes such as administrative, treatment, and clinical processes. Data intelligence capabilities allow ML, AI, and Deep learning to make the healthcare processes more accurate, fast and help the practitioner handle the increasing number of cases and processes. These advanced technologies help to extract real-time intelligence and make decisions regarding the diagnosis process, prescribing medicines, hospital management, laboratory, patient care, etc., and leads to high operational efficiency and care delivery.
• Supply chain management: Supply chain software generates and collects a vast amount of data. But they are not aware of how they can best use it to make their operations more effective. Data intelligence in the supply chain management network predicts business risk, minimizes loss, and makes automated self-learning supply chains. As a result, it drives real-time coordination and innovations.
• Human Resource: Organizations are using HR software to manage internal HR functions such as payroll, employee benefits, recruitment, training, talent management, attendance management, employee engagement, etc., to enhance their features and capabilities. They always have to do many tasks to understand employees better, attract top talent, and initiate programs to retain them and analyze their performance. They have a lot of data generated from their HRMS(Human Resource Management System) software. Data Intelligence can help them analyze and understand the data, gather insights, and make a precise decision that can make their organization drive healthier and faster.
• E-commerce: One of the success secrets of an e-commerce website is using customer reviews to know their experience, preference and then use them to make profitable decisions. Using ML and NLP techniques to interact with their customers and get data from them and use it to drive performance, improve Customer Engagement, Service Quality, Support Quality, and ultimately Sales. Data Intelligence makes it possible for them to accomplish these tasks, recommend products, understand customer preferences, solve their queries, improve quality and services, etc. Harnessing this information can give you a treasure trove of insights that can power your products and processes, improve customer experience, marketing, manage store operation, etc.

Conclusion

Akira AI is a data intelligence platform that provides intelligence using analysis and learning by processing data from various sources.

LF AI & Data Resources

• Learn about membership opportunities
• Explore the interactive landscape
• Check out our technical projects
• Join us at upcoming events
• Read the latest announcements on the blog
• Subscribe to the mailing lists
• Follow us on Twitter or LinkedIn
• Access other resources on LF AI & Data’s GitHub or Wiki