Navigating the Ethics of Artificial Intelligence:
Artificial Intelligence products like Siri and ChatGBT, have and continue to permeate both our personal and work lives. Further, AI usage is expected to dramatically increase over the coming decade as rapid product evolution leads to increased efficiency, cost savings, and innovation. However, this rapid rise comes with a multitude of ethical dilemmas, a topic that has become largely discussed among experts. Ethical considerations must be at the forefront of AI development to protect society.
Issues: Ethical Considerations
As noted above, AI is evolving quicker than the establishing of regulatory and ethical guidelines. Accordingly, several issues have arisen including (i) Accuracy (ii) Autonomy / Decision Making (iii) Bias & Discrimination (iv) Accountability and (v) Privacy.
Accuracy: Various AI models operate as “black boxes” and therefore produce content / decisions without clear rationale raising concerns surrounding accuracy. This was most evident in a court case in which ChatGBT made up various cases to substantiate a lawyer’s arguments.
Autonomy: AI is becoming increasingly sophisticated and independent. Although that “independence” is currently capped, we have to ask ourselves how much autonomy should be granted in the “delegation of decision-making power”. One area in which this will be particularly relevant is the army or the classic “trolley” problem.
Bias & Discrimination: In order to train AI, programmers use historical data that often includes biases. Accordingly, AI will often (i) perpetuate these biases or (ii) overcorrect these biases potentially leading to discriminatory hiring practices or other “bias” issues. Illustratively, in 2020 Michigan State Police arrested Robert Williams as a watch thief — their facial recognition tech, however, had identified the wrong guy.
Accountability: Whatever degree of autonomy is granted to AI begs the question: who or what is responsible for accountability in the event of mistakes. Is it the developer, user, or organization utilizing the product?
Privacy: AI usage will enhance existing concerns surrounding data collection and a decreasing amount of individual privacy. With more intelligent systems capable of gathering and efficiently analyzing data about individuals, we have to ask ourselves how much data should AI be privy to?
Implications:
The implications of non-ethical AI is vast. Without regulation, systems can become smarter than humans (as Elon Musker predicts) with unknown consequences. Individuals can abuse AI for more lewd purposes — like the deepfake of AOC in pornography. Nefarious actors will be able to easily spread disinformation by creating fake videos or content. Without clear ethical guidelines, we can live in a “post truth” society where it will be impossible to believe anything we don’t see with our own eyes. There is no end to the examples that can be used to demonstrate the dangers of unethical AI. The technology, unchecked, will likely have the power to affect every level of business & society. Just this week, Japan warned of this very issue. Accordingly, steps must be taken to ensure the standards remain ethical.
Solution:
Government & Industry leaders should institute clear & enforceable guidelines that prioritize accountability, and fairness. This will help the various “players” understand their exact roles & responsibilities. Further, in these negotiations, there must be international agreements & standards — without them, many of the risks outlined above will not be solved for (Note — solving for cross border data flow will be tricky if not universally agreed to). To preserve privacy, governments can use the European General Data Protection Regulations as inspiration. Allowing individuals to select what can and cannot be shared will be critical in ensuring individual data is not misused. Further, AI should be designed in a manner that provides clear rationale as to how it came to ABC decision. This will help reduce issues around accountability as the actions will be more clearly defined and help identify who is truly at fault. Finally, we suggest that universities and research organizations be incentivized to prioritize ethical AI research. This will help raise “ethical awareness” and ensure the next generation is able to navigate the complex ethical challenges that have and continue to arise with evolution of AI.
Conclusion:
AI is a powerful tool that can help reduce disparities in all areas of society and increase the standard of living for those that are able to utilize it. However, as noted earlier, it also poses major risks if not developed properly. Accordingly, we must be working on AI ethics at an even quicker pace than the development of AI to ensure that we preserve human values. To do so will require bipartisan support both domestically and internationally to ensure AI is developed in a manner with transparency, fairness, and accountability.
