In 2017, the Future of Life Institute developed important principles for AI development [now known as the 23 Asilomar Principles]. A year after, the Organization for Economic Cooperation and Development (OECD) held its own conference that centered on intelligent machines and how they can be utilized ethically. This article (attached below) talks about the 10 essential ingredients which can be used as a benchmark in judging whether an AI is ethical or not.
Here is the summary of the 10 principles of ethical AI according to OECD.
- AI must be transparent. As more and more AI is being utilized in workplaces, human employees must be provided with mechanisms to understand how decisions and outcomes are achieved. To achieve transparency, there must be consultation during implementation, development, and deployment.
- AI systems must be equipped with ethical black box-relevant data to ensure transparency and accountability.
- AI must remain compatible with the promotion of human dignity, integrity, freedom, cultural and gender diversity, and human rights. It must also serve to protect the planet.
- Adoption of a human-in-command approach. We must remember that AI is just a tool, and must not be given any other legal status that contradicts this understanding. In doing so, humans retain control and responsibility for the actions of these machines.
- AI should be genderless, and unbiased.
- Share the benefits of the AI system. Its benefits must be distributed broadly, and should be aimed at bridging the economic, social, and technological divide among peoples.
- Ensure that there is a just transition to AI. Aside from helping displaced workers find new employment, governments must provide safety nets that will continue to support fundamental freedoms and rights.
- Establish global governance mechanisms. Such governance must utilize a multi-stakeholder approach, involving AI designers, researchers, academics, employers, and workers, among others.
- Ban the attribution of responsibility to robots. This is connected with Principle #4 above, but with an additional provision which states that robots must be developed in accordance with existing laws protecting fundamental rights and freedoms.
- Ban AI arms race and the development of lethal autonomous weapons.
Why It Matters
In general, these 10 principles are good guidelines to follow mainly because it reiterates several deep-rooted concerns about AI implementation. It also stresses the fact that more than AI deployment, what we need to address are the current social, political, and economic concerns which plague our civilization for over a hundred years now. AI was imagined to be a solution to these problems, but as can be seen from this article, these conditions will make matters worse.
The current social reality shows that conflict is increasing in all levels. Countries continue to build weapons [see Pentagon Will Spend $2 Billion For AI Weapons and Kaist, Hanwa Systems Collaborate To Build AI Weapons], fundamental rights like privacy are being violated, and citizens find it hard to unite over issues thanks to the proliferation of fake news [Deep Fakes Are Worse Than Fake News].
While we believe that chaos can be a good thing as it marks the creation of new forms [see Who Is Ilya Prigogine?], we need to realize that our social structures are man-made. They do not follow the rules of nature, and as such, do not self-heal. If chaos becomes deep-seated in our societies, it could lead to our total destruction. If we continue with the same kind of thinking that has built our current civilization (materialism), then we can expect that we would annihilate ourselves long before nature has had the chance to implode (due to climate change).
The good news is that the new sciences give us a new hope for the future. In the same way that we have been awakened to become better people, have used the power of our thinking to change our personal realities, we must also dream, advocate, and act towards creating our societies into better communities. New generation of humans intuitively understands the need for a just, equal, and fair society. Many of them have the skills to actually make this happen. But they are slowly being spoiled by the education they have to undergo – an education that was designed for societies 100 years ago.
So much needs to change in our current society, if we are to create a better one. AI will be a tool that will help us achieve future goals. But we must, as early as now, redirect our efforts. We must change our mindsets, and ultimately, our institutions. It sounds cumbersome, but with new science, there is hope.Ethical-artificial-intelligence-10-essential-ingredients