The Question of AI Usage
In this article, Judaea tackles the question of how AI should be used in a rapidly advancing world.
Embedded Files
How Should AI be Utilized?
How Should AI be Utilized?
By: Judaea Ramos
By: Judaea Ramos
In a rapidly advancing world, AI and robotic machinery are quickly becoming a regular, everyday usage in every aspect of life. Of course, ethical issues have revolved around AI and whether replacing human people with machinery is a good idea or not. This is a heavy issue for some people, as many companies and businesses have started actually implementing AI into more highly specialized fields, boasting results that, at best, vary greatly. While it seems that everyone is arguing when and where AI should be used, they all seem to be missing a greater, more important question: how should AI be used? This can easily be answered with a simple statement: it can be viewed as a tool, like one would use a drill or a hammer to build houses.
Image from IT Chronicles
To start thinking about this, it is important to get out of the ‘AI is better than humans at anything!’ mindset, as well as the stiff ‘AI can’t replace real human such and such!’ train of thought. As touching as the latter statement may be, it really does not answer the question of how we should use this new technology; it more or less seems to stem from a fear of replacement, or a fear of the technology actually being better than we are. Tom Hoopes, a writer for Benedictine College, brings up another book by Henry Kissinger and his co-authors, called The Age of AI. In it, Tom says that Google’s DeepThink AI, named AlphaZero, was put up against Stockfish, an old computerized chess opponent, coded with all the inputs of the best human strategies in the game. AlphaZero, on the other hand, had only a few hours of experience; the program was quite literally thrown into the fire to see what would happen.
In a (quite terrifying) display of machine learning, Tom describes how AlphaZero — having had four hours to train with just the basic rules and objectives of the game of chess — “beat Stockfish 155 games to 6, with 1 draw.” It did this by sacrificing “its own most precious pieces — including its queen — to move in on its enemy with a cold efficiency greater than any human mind ever conceived.” While this is a strange way of playing the centuries-old game, it does raise an interesting question: does AI simply make up for the mistakes and missed opportunities that humans make all the time? Does it show us all the opportunities that are actually available to us, and demonstrate to us that we can actually learn from this creation of ours? When and how shall it be implemented?
Image by Tenor
IBM’s training manual from 1979 fundamentally sums up AI’s implementation. In an excerpt that has grown in popularity, the manual states, “A computer can never be held accountable. Therefore, a computer must never make a management decision.” Ultimately, the computer does what it is programmed to do, so all the fault lies on the programmers and the decision-makers who designed the computer to try and make it do all the managerial decisions for them.
Image by Harvard Business Review
This, interestingly, puts the blame on human error, rather than machine miscalculation. As the use of AI increases rapidly, with the expansion of such technology being profound, it would be best for humanity to realize that they have total control over this technology. If they were to change the popular mindset that this tool will take their job because it ‘does it better’, then perhaps more advancements could be made. This is precisely what AI is and should be: a tool that factors in and makes up for human failure.
Most predominantly, the rapid rise of AI has been linked to a rapid rise of ethical concerns regarding the technology. According to an article by the United Nations Educational, Scientific and Cultural Organization (UNESCO), four core values can make this technology benefit the greater good of humanity and work well with individuals, societies, and the environment. These include:
“Human rights and human dignity.” This is so that AI may respect, protect, and promote human rights, fundamental freedoms, and dignity.
“Living in peaceful, just and interconnected societies.”
“Ensuring diversity and inclusiveness.”
“Environment and ecosystem flourishing.” This is most definitely a regulatory act that protects wildlife and habitats as a whole.
Additionally, UNESCO proposes ten core values/principles that follow a human rights-centered approach to working with AI:
“Proportionality and Do No Harm.” This is so that the use of AI does not go beyond ethical limitations.
“Safety and Security.” Unintentional damage and other harms dealt by AI actors should be addressed by them.
“Right to Privacy and Data Protection.” This ensures that privacy is protected and promoted.
“Multi-Stakeholder and Adaptive Governance & Collaboration.” This makes sure that data usage does not violate international law or national sovereignty, as well as motivates diverse stakeholders to search for inclusivity in AI governance.
“Responsibility and Accountability.” According to this UNESCO principle, “AI systems should be auditable and traceable,” and demand appropriate oversight of the technology, assessments on the impacts of the technology, and the establishment of mechanisms that will not conflict with “human rights norms”, as well as not posing any threats to the environment.
“Transparency and Explainability.” This principle is entirely dependent on context, but it does ensure that AI systems to be deployed are, as stated, transparent and explainable.
“Human Oversight and Determination.” Possibly the most important, as it does refer back to the IBM excerpt, this determines that AI does not misplace “ultimate human responsibility and accountability.”
“Sustainability.” AI is to be assessed against the “constantly evolving goals” so that its impacts on sustainability can be lessened (to the extent that can be done for this confusing principle).
“Awareness and Understanding.” Public knowledge of AI systems/data should be accessible to all through multiple forms.
“Fairness and Non-Discrimination.” This forces AI actors to make certain that their systems are non-discriminatory and inclusive, promoting social justice and allowing everyone to reap the benefits of these systems.
These principles are set to guide actors on how these systems and datasets should be handled. In plain terms, some of the most stressed principles are the avoidance of violations against human rights, coexisting peacefully with society, the environment, and the idea of not preferring one party over the other. In a clear, concise manner, UNESCO details how AI should be used, as well as how not to use it. For instance, it can be used to endanger the safety of humanity, as well as to compromise the security of information and data, which is something AI has access to at all times.
As the world continues to advance in terms of technology, a proper limiter in this advancement is fear and misunderstanding of the tool humanity has in its hands. AI is fully under the thumb of humanity but, even so, it is a powerful tool that has and continues to revolutionize in almost every field in which the system is implemented. Humans control the tool and, again, it has always been humans who shape their own lives.
If the wielder is afraid of the tool he made, then why would he have made it in the first place?
Page updated
Report abuse