This resource introduces the topic of information privacy and AI to a broader audience. It is not intended to answer queries or offer legal advice because it is written for a non-technical audience. It should be mentioned that this paper does not cover all of the ethical, technical, and legal concerns related to Artificial Intelligence.
Describe artificial intelligence (AI) at a high level, discuss its applications in the public sector, and point out some of the potential and problems it raises for data protection.
In its most basic form, artificial intelligence (AI) is a branch of computer science that aims to develop systems that are capable of carrying out tasks that are typically completed by humans. These tasks, which involve visual and auditory perception, learning and adaptation, reasoning, pattern recognition, and decision making, might be regarded as intelligent. A group of related methods and technologies, such as robots, machine learning, predictive analytics, and natural language processing, are collectively referred to as artificial intelligence (AI).
Although many people are unaware of it, Artificial Intelligence technologies are already being used in real-world situations in our daily lives. One of AI’s features is that, once the technology is functional, it ceases to be called AI and becomes a part of everyday computing.2. Examples of commonplace AI technology include hearing an automated voice on the other end of the phone or having a movie recommendation based on your tastes. Since these technologies are already a commonplace part of our daily lives, people frequently forget that artificial intelligence (AI) techniques like speech recognition, natural language processing, and predictive analytics are in use.
Artificial intelligence that is limited, general, and super
The majority of Artificial Intelligence that exists today is “narrow.” This indicates that it has been specifically programmed to be proficient in a single field. To emphasize its capacity to supplement (rather than completely replace) human intelligence, it is also occasionally referred to as augmented intelligence. For instance, IBM’s Deep Blue computer, created in the 1980s, is capable of playing chess at a level higher than humans; this achievement is crucial to the history of artificial intelligence development. Deep Blue’s intellect stops there, even though it demonstrates superior chess skills.
On the other hand, the term artificial general intelligence (AGI) describes a degree of intelligence that spans several domains. The natural world already demonstrates the difference between narrow and general intelligence: ants and bees are instances of narrow intelligence, as they both know how to construct a nest and beehives, respectively.
But this intelligence is domain-specific; ants cannot construct a hive, and bees cannot construct a nest. In contrast, humans possess the ability to be intelligent in a variety of domains and can pick acquired intelligence in new ones through observation and experience.
Big Data
AI and big data have a mutually beneficial relationship. Although big data analytics procedures are currently in place, artificial intelligence (AI) approaches are necessary to fully realize the potential of big data. On the other hand, big data provides Artificial Intelligence with a vast and abundant stream of input data for growth and learning. In this way, big data and artificial intelligence are closely related.
Although the term “big data” has no universally accepted definition, it is typically used to refer to vast volumes of data generated and gathered in many different formats.5. It is impossible to overstate the types and volume of information that fall under the umbrella of “big data.” Almost everything people do, whether on purpose or accidentally, produces a great deal of data about them, including searching the internet, exchanging and exchanging daily information with the government, businesses, and social media, and simply using a smartphone while out and about. The amount of data generated, gathered, and fed into Artificial Intelligence systems has the potential to expand into our personal lives as the Internet of Things (IoT) drives the network deeper into our physical surroundings and private areas.
Machine Learning
A method in computer science called machine learning enables computers to “learn” on their own. Although Artificial Intelligence is frequently used to describe it, that is just one aspect of it. Machine learning’s dynamic capacity to adapt to new data sets it apart from other types of artificial intelligence. The machine is training itself by consuming data and creating its own reasoning based on the data it has examined.
Supervised and unsupervised machine learning are the two primary categories. A human must supply the data and the answer for supervised learning, allowing the machine to discover how the two are related. By consuming a lot of data—typically huge data—and repeatedly going through it to identify trends and insights, unsupervised learning gives the machine more freedom to learn.
These methods are applied in various settings and for various objectives. Because neither requires explicit programming on what to look for, the system has some autonomy to come up with its own reasoning and spot patterns that humans might have missed otherwise.7. In today’s world, machine learning algorithms are already widely used. Creating web search results, suggesting services like Netflix and Pandora, and estimating a product’s monetary value based on the current market are a few examples. The input data that is supplied determines how beneficial machine learning is. Consequently, huge data has been essential to machine learning’s success.
Most frequently used to describe deep neural networks, deep learning is a subset of machine learning.8. In general, a neural network uses a layered approach to process data, with each layer receiving its input from the one that came before it. The neural network’s layer count is referred to as “deep.”
It can get harder to fully understand the choices and deductions made at each level as the product of one layer becomes the input of the next. It can be difficult to fully comprehend and explain the actions that lead to a specific outcome when moving through each layer because of what is known as the “black box” effect.9. Neural networks are sometimes explained using the human brain as an analogy, although this is not very useful because it suggests that robots comprehend information similarly to how humans do, which is untrue.
Artificial intelligence in the public sector
Although industry and university research are the primary forces behind the development of Artificial Intelligence technology, the public sector can also benefit from Artificial Intelligence applications and advancements. Although AI is already used in many government functions, more widespread usage of these technologies could be advantageous. Furthermore, by enacting laws, establishing policies, and showcasing best practices, the government may significantly influence how AI technology affects people’s lives. It is crucial that the government keeps up with the private sector’s rapid advancement; this calls for a proactive, dynamic, and knowledgeable approach to technology and how it interacts with society and the law.
Resources, technological prowess, and public trust continue to be the limiting factors for AI’s present and future use in government. AI’s ability to lessen administrative workloads and assist in resolving resource allocation issues is among the public sector’s greatest immediate advantages. AI applications have the potential to significantly improve the efficiency of long-standing government processes in the near future, including query response, document completion and search, request routing, translation, and document authoring.11. For instance, some of the bigger Australian government agencies now employ chatbots to offer guidance and customer support to people.
AI has the ability to completely transform government operations in the long run, going beyond simply improving current procedures. Organizations will probably need to adjust to the changing demands and expectations of citizens as well as change the legal and regulatory environment to accommodate new technological applications.
Although AI has great promise for the public sector, it cannot be viewed as a solution to every problem facing the government today. Strategic and careful implementation of AI technology use and regulation is required, with special attention paid to information management, including privacy, protecting data security, and ethics in general.Twelve
Privacy considerations
Some of the most important concerns raised by AI in regard to information privacy are examined in this section. This is meant to serve as a summary and a starting point for additional conversation on some of the more important information privacy concerns; it is not a comprehensive examination of every topic.
The 1980 OECD Guidelines on the Protection of Privacy and Transborder Flows of Personal Data serve as the fundamental foundation for information privacy law in Victoria and elsewhere. Eight fundamental concepts found in these guidelines are still upheld by privacy laws worldwide, such as the Privacy and Data Protection Act 2014 (PDP Act). One advantage of principle-based regulation is that it acknowledges the complicated and multifaceted nature of privacy and permits some latitude in how privacy can be safeguarded in different situations, along with changing social norms and technological advancements. Although the OECD Guidelines have been extremely effective in advancing information privacy laws globally, artificial intelligence poses threats to the fundamental ideas that underpin the Guidelines.
Even though AI could challenge conventional ideas of privacy, this does not mean that privacy will always be compromised; in fact, it is feasible to think of a future when privacy will be enhanced by AI. In order to deal with raw data, for example, fewer individuals will likely require access to it, which could reduce the possibility of human error-related privacy violations. Additionally, it might enable more meaningful consent, whereby people receive tailored services based on their gradually learned privacy preferences. The current state of privacy protection may need to be reviewed due to the growing usage of AI, but this does not imply that privacy will vanish or become obsolete.
Why is AI different?
Important privacy concerns are generally always present in emerging technologies, but the scope and use of AI present a special and unprecedented set of difficulties. In certain respects, the implications of artificial intelligence (AI) can be viewed as an extension of those generated by big data. However, AI technology offers not only the capacity to process vast quantities of data, but also the capacity to learn from it, create adaptive models, and generate actionable predictions—much of which is accomplished without clear, explicable procedures.
As AI technology advances, there is a serious chance that the presumptions and prejudices of the people and organizations developing it will affect the AI’s results. Government agencies that want to use neural networks for decision-making face difficulties due to the unintended consequences brought on by biases and the opaque outcomes of this technology. The potential for prejudice and its relationship to privacy are covered in more detail below.
Additionally, AI can alter how people communicate with technology. Many forms of AI already exhibit human traits, for example. Human-like interfaces, such the human-sounding voices seen in assistants like Alexa and Siri, may give rise to new privacy issues. Research from the social sciences shows that individuals tend to treat technology like a human.13 With AI that mimics human traits, people may be more likely to build trustworthy relationships and, as a result, be more willing to divulge more personal information than with other types of technology that gather data in a more conventional way.
The growing power gap between the organizations that collect data and the people who create it has not been taken into consideration in most of the information privacy rhetoric surrounding AI.14 Current models typically treat data as a commodity that can be exchanged, which fails to fully recognize how difficult it is for people to make decisions about their data when interacting with systems they don’t fully understand, especially when the system knows them well and has learned how to manipulate their preferences by ingesting their data. Furthermore, a lot of AI’s adaptive algorithms are always evolving, to the point that their developers frequently find it difficult to adequately describe the outcomes they produce.
