The Rise of Automated Decisions in Modern Life

Automated decisions are becoming a norm in our daily lives, and it's essential to understand their impact. Many of us have experienced automated decisions firsthand, such as receiving a flight delay notification on our phone.

These notifications are often made possible by algorithms that analyze data in real-time to predict and prevent disruptions. For instance, airlines use machine learning algorithms to identify potential delays and send notifications to passengers.

In recent years, the use of automated decisions has increased exponentially, with many industries adopting this technology to streamline processes and improve efficiency. According to a study, the global automated decision-making market is expected to grow by 25% in the next five years.

As we continue to rely on automated decisions, it's crucial to consider their limitations and potential consequences.

Automated decisions are becoming increasingly common, and it's essential to understand what they're all about. ADM, or Automated Decision-Making, is a broad term that encompasses a range of technologies and applications.

There are different definitions of ADM, but one thing is clear - it involves decisions made through technological means, often without human input. The EU's General Data Protection Regulation (Article 22) suggests that ADM involves decisions made solely through technology.

ADM can take many forms, from decision-support systems that make recommendations for human decision-makers to fully automated decision-making processes that make decisions on behalf of individuals or organizations. This can be as simple as checklists and decision trees or as complex as artificial intelligence and deep neural networks (DNN).

Computers have come a long way since the 1950s, from basic processing to complex tasks like image and speech recognition, gameplay, scientific, and medical analysis. This increased capacity has enabled ADM to be deployed across various sectors and domains.

An ADM system (ADMS) can involve multiple decision points, data sets, and technologies (ADMT), and may sit within a larger administrative or technical system. This can include systems like a criminal justice system or business process.

Automated decision-making technologies (ADMTs) are software-coded digital tools that automate the translation of input data to output data, contributing to the function of automated decision-making systems.

ADMTs involve a wide range of technologies, including basic computational operations like search, matching, and mathematical calculations. These operations can be used for tasks such as 1-2-1, 1-2-many, and data matching/merge, which are essential for any automated decision-making system.

User profiling is another important aspect of ADMTs, which involves creating a detailed picture of a user's behavior, preferences, and characteristics. This information can be used to create personalized recommendations, which is a key feature of recommender systems.

Clustering and classification are also common ADMTs that involve grouping similar data points together and categorizing them based on certain criteria. Feature learning is another important technique that involves identifying the most relevant features of a dataset.

Predictive analytics, which includes forecasting, is a powerful ADMT that involves using statistical models to make predictions about future events or trends. This can be used in a variety of applications, from weather forecasting to financial modeling.

Social network analysis, which includes link prediction, is another ADMT that involves studying the relationships between individuals or groups. This can be used to identify patterns and trends in social networks, which can be useful for applications such as marketing and public health.

ADMTs can also be used to process complex data formats, such as images, audio, and natural language. Image processing involves using algorithms to analyze and manipulate images, while audio processing involves using algorithms to analyze and manipulate audio signals.

Business rules management systems are another type of ADMT that involves using software to manage and enforce business rules and regulations. Time series analysis involves using statistical models to analyze and forecast data that varies over time.

Anomaly detection involves using ADMTs to identify unusual patterns or outliers in a dataset, which can be useful for applications such as fraud detection and quality control. Modelling and simulation involve using ADMTs to create virtual models of real-world systems, which can be used to test and optimize complex systems.

Here are some examples of ADMTs and their applications:

Data and quality are crucial factors in automated decision-making systems. The quality of available data is often highly problematic due to its variability.

Datasets can be incomplete, biased, or limited in terms of time or coverage. They may also use different measuring and describing terms. For instance, datasets used in ADM systems might be restricted for privacy or security reasons, making it difficult to obtain or compute large corpora.

Large corpora are often required for machines to learn from data, which can be challenging to obtain or compute. However, where available, they have provided significant breakthroughs, such as in diagnosing chest X-rays.

Data quality is a major concern when it comes to using data in automated decision-making (ADM) systems. The quality of the available data and its ability to be used in ADM systems is fundamental to the outcomes.

Datasets are often highly variable, and corporations or governments may control large-scale data, restricted for privacy or security reasons, incomplete, biased, limited in terms of time or coverage, measuring and describing terms in different ways, and many other issues.

For machines to learn from data, large corpora are often required, which can be challenging to obtain or compute; however, where available, they have provided significant breakthroughs, for example, in diagnosing chest X-rays.

A significant application of ADM in social services relates to the use of predictive analytics – eg predictions of risks to children from abuse/neglect in child protection, predictions of recidivism or crime in policing and criminal justice, predictions of welfare/tax fraud in compliance systems, predictions of long term unemployment in employment services.

The quality of the data used in these applications can have a significant impact on the outcomes. For example, Australia's federal social security delivery agency, Centrelink, developed and implemented an automated process for detecting and collecting debt which led to many cases of wrongful debt collection in what became known as the RoboDebt scheme.

In many cases, the data used in ADM systems is sourced from various places, including sensor data for self-driving cars and robotics, identity data for security systems, demographic and financial data for public administration, medical records in health, and criminal records in law. This can sometimes involve vast amounts of data and computing power.

Here are some common issues with data quality:

These issues can have a significant impact on the outcomes of ADM systems, making it essential to ensure that the data used is accurate and reliable.

Advertising is a complex and often automated process, especially in the digital world. Online advertising is closely integrated with many digital media platforms and websites, and often involves automated delivery of display advertisements in diverse formats.

Programmatic online advertising automates the sale and delivery of digital advertising on websites and platforms via software. This approach is sometimes known as the waterfall model, which involves a sequence of steps across various systems and players.

Internet users who dislike ads have adopted counter measures such as ad blocking technologies. In 2017, 24% of Australian internet users had ad blockers.

Lack of transparency for advertisers and unverifiable metrics are just two of the issues with this system.

Deep learning AI image models are being used for reviewing x-rays and detecting the eye condition macular degeneration. This technology has the potential to save countless lives by catching health issues early on.

Medical professionals are utilizing AI to analyze medical images, allowing them to make more accurate diagnoses.

Machine learning is a powerful technology that involves training computer programs through exposure to large data sets and examples to learn from experience and solve problems. This has led to breakthroughs in image and speech recognition, translations, text, data, and simulations.

Machine learning systems based on foundation models run on deep neural networks and use pattern matching to train a single huge system on large amounts of general data such as text and images. This has enabled the creation of programs like Open AI's DALL-E, an image creation program, and their various GPT language models.

Recent advancements in training deep neural networks and increases in data storage capacity and computational power have made machine learning increasingly powerful. This has led to the development of dynamic products like InRule's AI-powered business rules engine, which infuses AI throughout organizations.

Over 500 companies, including Allstate and Bank of America, rely on InRule Technology's accessible and explainable AI solutions. These solutions promote accuracy, customer satisfaction, and employee retention, among other benefits.

Machine learning has been around for some time, but recent breakthroughs have made it a game-changer in various industries. It's no wonder that companies are turning to machine learning to improve their operational efficiency and decision-making processes.

As automated decisions become more prevalent, it's essential to consider the debate and concerns surrounding their use. Research and development are underway to assess argument quality and evaluate conversational, mathematical, scientific, interpretive, legal, and political argumentation and debate.

The European Data Protection Board has highlighted concerns related to bias, lack of transparency, and accountability in automated decision-making systems. These concerns are crucial to address as these systems increasingly influence critical areas of our lives.

Legislative responses to automated decision-making include the European General Data Protection Regulation (GDPR), which enshrines the right of data subjects not to be subject to decisions based solely on automatic individual decision making. The GDPR also includes some rules on the right to explanation, although the exact scope and nature of these is currently subject to pending review by the Court of Justice of the European Union.

Information asymmetry can make it difficult for individuals to understand how decisions are being made about their data. Automated decision-making can increase the gap between those who provide the data and those who use it to make decisions.

Information asymmetry can be a significant issue in various fields, including financial trading. In fact, it's been observed that the information asymmetry between two artificial intelligent agents in financial trading is often less than between two human agents or between human and machine agents.

This disparity in knowledge can lead to unequal power dynamics and may even disadvantage individuals who are not aware of how their data is being used.

Research and development are underway into uses of technology to assess argument quality, assess argumentative essays and judge debates.

Potential applications of these argument technologies span education and society.

Scenarios to consider in these regards include those involving the assessment and evaluation of conversational argumentation.

Mathematical argumentation is also being explored through these technologies.

Scientific argumentation is another area where these technologies are being developed.

Interpretive, legal, and political argumentation and debate are also being assessed and evaluated using these technologies.

Concerns about ADM are multifaceted and far-reaching.

ADM may incorporate algorithmic bias arising from data sources, technical design of the algorithm, and emergent bias.

The European Data Protection Board highlights the need for a balanced approach to harness the benefits of ADM while addressing its potential risks.

Data sources can be biased in their collection or selection, which can lead to biased outcomes. Technical design of the algorithm can also introduce assumptions about human behavior, further exacerbating bias.

Emergent bias occurs when ADM is applied in unanticipated circumstances, creating biased outcomes.

Concerns raised about ADM include lack of transparency, incursions on privacy and surveillance, and exacerbating systemic bias and inequality.

ADM systems are often based on machine learning and algorithms that are not easily viewable or analysable, leading to concerns about their transparency and accountability.

A report from Citizen lab in Canada argues for a critical human rights analysis of ADM to ensure it doesn't infringe on rights, including equality and non-discrimination.

The European General Data Protection Regulation (GDPR) enshrines the right of data subjects not to be subject to decisions based solely on automatic individual decision making.

Some legislative responses to ADM include:

Automated decisions are being used in various industries to improve efficiency and accuracy. In finance, automated decisions are being used to detect and prevent money laundering, with some systems able to analyze thousands of transactions in a matter of seconds.

These systems are also being used in healthcare to help doctors make more accurate diagnoses. For example, a study found that AI-powered systems were able to accurately diagnose breast cancer 99% of the time, compared to 87% for human doctors.

Automated decisions are also being used in customer service to provide faster and more personalized responses to customers.

In the world of medicine, machine learning algorithms have been used to analyze medical images and detect diseases such as breast cancer.

The algorithm can identify tumors and distinguish them from healthy tissue with a high degree of accuracy.

The use of machine learning in medical imaging has been shown to improve diagnosis rates and reduce false positives.

This technology has been applied in hospitals and clinics around the world, including the University of California, Los Angeles (UCLA).

Researchers at UCLA have used machine learning to develop a system that can detect breast cancer from mammography images with 97% accuracy.

The system uses a combination of machine learning algorithms and deep learning techniques to analyze the images.

This technology has the potential to revolutionize the way doctors diagnose and treat breast cancer.

In the field of finance, machine learning algorithms have been used to detect credit card fraud.

These algorithms can analyze large amounts of data and identify patterns that may indicate fraudulent activity.

The use of machine learning in credit card fraud detection has been shown to reduce the number of false positives and improve the accuracy of detection.

This technology has been applied in banks and financial institutions around the world, including JPMorgan Chase.

JPMorgan Chase has used machine learning to develop a system that can detect credit card fraud with a high degree of accuracy.

The system uses a combination of machine learning algorithms and statistical models to analyze transaction data.

This technology has the potential to save banks and financial institutions millions of dollars in lost revenue each year.

In the field of transportation, machine learning algorithms have been used to optimize traffic flow and reduce congestion.

These algorithms can analyze real-time traffic data and adjust traffic signal timing to minimize congestion.

The use of machine learning in traffic optimization has been shown to reduce travel times and improve air quality.

This technology has been applied in cities around the world, including Singapore.

Singapore has used machine learning to develop a system that can optimize traffic flow and reduce congestion.

The system uses a combination of machine learning algorithms and data from sensors and cameras to analyze traffic patterns.

This technology has the potential to make cities more livable and reduce the environmental impact of transportation.

In the media and entertainment industry, recommender systems are increasingly being used to provide content to audiences. These systems use demographic information, previous selections, collaborative filtering, or content-based filtering to suggest content.

Digital media platforms, such as music and video platforms, publishing, and health information websites, are using these recommender systems to offer personalized content to users. This approach has been successful in increasing user engagement and satisfaction.

Large-scale machine learning language models are being developed by companies like OpenAI and Google, which have the potential to revolutionize fields such as advertising, copywriting, and graphic design. These models can generate high-quality content quickly and efficiently.

Automated content creation, such as image creation programs, are also being developed and are likely to have widespread application in fields like journalism and law.

In the transport and mobility sector, connected and automated mobility (CAM) is revolutionizing the way we travel. CAM involves autonomous vehicles like self-driving cars that use automated decision-making systems to replace human control.

Cars with levels 1 to 3 of automation are already available on the market in 2021. These vehicles can perform tasks like steering, acceleration, and braking, but still require human intervention.

The German government established an 'Ethics Commission on Automated and Connected Driving' in 2016, which recommended developing CAVs if they cause fewer accidents than human drivers. This commission also provided 20 ethical rules for adapting automated and connected driving.

The European Commission strategy on CAMs in 2020 aimed to reduce road fatalities and lower emissions by adopting CAVs in Europe. However, self-driving cars also raise concerns about liability, ethical decision-making, and privacy issues.

As we move towards a future with more autonomous vehicles, addressing issues of trust and safety will be crucial for their widespread adoption.

Automating decisions can bring numerous benefits to businesses. Increased efficiency is one of the primary advantages, allowing for faster responses to customer needs and market changes.

Automating decision-making processes can significantly reduce the time taken to reach conclusions. This enables businesses to respond quickly to changing circumstances, giving them a competitive edge.

Consistency and accuracy are also key benefits of automated decisions. By relying on predefined rules, organizations can ensure that decisions are consistent across the board, reducing the likelihood of errors and biases.

Automated decision-making can handle complex datasets, providing insights and accuracy beyond human capabilities. This is particularly useful in industries where data analysis is crucial, such as healthcare and finance.

Businesses can also benefit from automated decisions in terms of scalability and cost reduction. Rules engines allow for easy updates and modifications to business rules, offering the agility needed to adapt to changing business environments.

Here are some examples of how automation can deliver value in different areas:

Automating routine decisions can also improve operational efficiency, allowing staff to focus on more strategic and value-adding activities. Business analysts can benefit from faster development cycles and reduced maintenance costs, while IT leaders can enjoy scalability and interoperability.