"There are insufficient mechanisms to foresee and measure negative impacts, andto promote and safeguard positive impacts of A/IS.
## BackgroundA/IS technologies present great opportunity for positive change in every aspect of society. However, they can—by design or unintentionally— cause harm as well. While it is important to consider and make sense of possible benefits, harms, and trade-offs, it is extremely challengingto foresee all of the relevant, direct, andsecondary impacts.However, it is prudent to review case studies of similar products and the impacts they have hadon well-being, as well as to consider possibletypes of impacts that could apply. Issues to consider include:
Economic and labor impacts, including labor displacement, unemployment, and inequality,
Accountability, transparency, and explainability,
Surveillance, privacy, and civil liberties,
Fairness, ethics, and human rights,
Political manipulation, deception, “nudging”,and propaganda,
Human physical and psychological health,
Environmental impacts,
Human dignity The human right to be valued and treated with respect because of one's personhood. , autonomy, and human vs.A/IS roles,
Security, cybersecurity, and autonomous weapons, and
Existential risk and super intelligence.While this is a partial list, it is important to be aware of and reflect on possible and actual cases. For example:
A prominent concern related to A/IS is oflabor displacement and economic and social impacts at an individual and a systems level.A/IS technologies designed to replicate human tasks, behavior, or emotion have the potential to increase or decrease human well-being. These systems could complement human work and increase productivity, wages, and leisure time; or they could be used to supplement and displace human workers, leading to unemployment, inequality, and social strife.It is important for A/IS creators to think about possible uses of their technology and whether they want to encourage or design in restrictions in light of these impacts.
Another example relates to manipulation. Sophisticated manipulative technologies utilizing A/IS can restrict the fundamental freedom of human choice by manipulating humans who consume content without them recognizing the extent of the manipulation. Software platforms are moving from targeting and customizing content to much more powerful and potentially harmful “persuasive computing” that leverages psychological data and methods. While these approaches may be effective in encouraging use of a product, they may come at significant psychological and social costs.
A/IS may deceive and harm humans by posing as humans. With the increased ability of artificial systems to meet the Turing test, an intelligence test for a computer that allows a human to distinguish human intelligence from artificial intelligence, there is a significant risk that unscrupulous operators will abuse the technology for unethical commercial or outright criminal purposes. Without taking action to prevent it, it is highly conceivable that A/IS will be used to deceive humans by pretending to be another human being in a plethora of situations and via multiple mediums.A potential entry point for exploring these unintended consequences is computational sustainability.Computational-Sustainability.org defines the term as an “interdisciplinary field that aims to apply techniques from computer science, information science, operations research, applied mathematics, and statistics for balancing environmental, economic, and societal needs for sustainable development”. The Institute of Computational Sustainability states that the intent of computational sustainability is provide “computational models for a sustainable environment, economy, and society”. Examples of applied computational sustainability can be seen in the[ Stanford University Engineering Department’s course in computational sustainability presentation](https://cs.stanford.edu/~ermon/cs325/slides/lecture1-S
pdf). Computational sustainability technologies designed to increase social good could also be tied to existing well-being metrics."p.83-86