page image

CHAPTER IV - Identifying Potential Harms

Smart regulation of digital and network technologies requires solid understanding of both benefits and risks. Before moving to solutions, conferees brainstormed about potential harms associated with new technologies, such as AI, 5G and IoT. These harms include:

Misuse of personal data/loss of privacy. Many individuals feel they have lost control over their personal data. Some may be willing to give up personal data in exchange for free services, but others are suspicious of how their data are used. The Pew Research Center has found that a majority of Americans are wary of computer algorithms being used to make decisions in areas with a real-life impact on their lives, such as resume screening, job interview evaluations, personal finance scoring and assessments of criminals up for parole. For many, the lack of transparency and control creates anxiety. Some are alarmed by the potential loss of personal freedom and manipulation stemming from data collection.

Algorithmic bias. Artificial intelligence is only as good as the algorithms that power the underlying decision-making. Algorithms are trained to make inferences from massive data sets, aided by machine learning. The risk is that AI may have implicit biases through either coding decisions made by engineers or as a result the specific datasets used to create the algorithm. For instance, the failure to include data from marginalized communities can lead to profiling and stereotyping regarding products and services. At the same time, others are concerned about the potential harm to innovation if we overly constrain technologies, like AI.

Warrantless government surveillance. Surveillance is increasingly common, both by governmental agencies and private entities. An emerging issue is how that information is being used. Facial recognition technology is particularly an area of concern. It has been reported that the Federal Bureau of Investigation is scanning photos from state Department of Motor Vehicle databases, and several cities have banned the use of facial recognition software by local law enforcement. Critics fear of a loss of privacy, unchecked surveillance and a chilling of First Amendment rights.

Misinformation. People do not agree on which source of information they can trust. Individuals lack the ability to validate information that is presented as a fact. Content can be digitally altered, potentially causing great harm. Misinformation on online platforms can spread and quickly go viral.

Harmful content/extremist speech on online platforms. Harmful and extreme speech can be amplified online, reaching a much larger audience. Algorithms sometimes drive users towards more extremes. Technology companies are struggling to respond.

Violations of civil rights. Existing laws prohibit discrimination, but AI techniques bring new challenges to the forefront. Machine learning may be used to discriminate against protected groups—even if that is not the intent of engineers designing the algorithms—by targeting advertisements to groups with certain characteristics. Discrimination laws may need to be modernized for the digital age, or existing laws may require new enforcement mechanisms.

Cyber risk. Digital communications provide new opportunities for fraudulent activity ranging from extortion to malicious actions to the complete shutdown of critical systems. Fraudsters create virtual personalities to extract benefits from unsophisticated or trusting individuals. IoT exponentially increases the number of points of attack. In a connected world, anything can be subject to a crippling cyber-attack—banks, utilities, hospitals, stores, offices and even entire cities.

Potential loss of U.S. technological leadership. The U.S. led the world with the advent of the smart phone and the creation of 4G applications. The rest of the world took note, and now other nations want to lead in 5G. The potential loss of U.S. technological leadership represents a significant threat to American consumer choice, innovation, competition and security. A pervasive concern of the U.S. is the threat of China supremacy in these areas.

Threats to U.S. entrepreneurship. Some individuals view the dominance of the largest U.S. technology firms as a threat to continued inno-vation and entrepreneurship. Absent a continuous cycle of new entrants, some people fear the loss of the benefits of disruptive innovation.

Digital divide. The digital divide is a multi-dimensional problem that harms individuals, communities and society at large. It includes problems of broadband access, affordability and adoption resulting in the inability of individuals to fully participate in the digital economy. Specific harms stemming from digital inequality include reduced civic participation and limited access to employment opportunities, healthcare benefits, educational opportunities, commerce and social media.

Uneven or disproportionate distribution of benefits that come from technological advances. When portions of the population remain unconnected, there are significant, unquantified costs to society at large. The challenge for policymakers is to devise effective interventions to address these harms without stifling the benefits. Policymakers need to be mindful that by taking action to prevent some of these harms, they may create new, even more serious problems.

Title Goes Here
Close [X]