Future of Ethical AI: The best way to predict your future is to create it.
Georg Olowson & Sophia Greulich - Advocates for Trustworthy AI, Consultants and Optimists
Even if you don't realize it, you interact with artificial intelligence every day. For example by communicating directly with a voice assistant, in social media, or hidden in a process. We expect it to make us more efficient, faster and more informed - it should make our lives easier and give us more time for the important things. But so much for the ideal world.
AI can be very efficient and augment people in their work. We are good at putting AI to work. But are we putting good AI to work??
If AI can help doctors make more informed decisions and be on top of the latest research, that is great. If AI can help companies secure data or predict future outcomes, that is cool and helpful.?If AI can help police identify you at any point in time and knows everything you searched, wrote about and what your secrets are, that is .... wait a second.?
With AI, as with any other tool, it depends on what purpose, in what context, in what way, and with what impact it is used. Only then can you weigh whether it is also a good use.?We will come to a better understanding of what is good.
If AI leads to biased decisions, for example, women getting worse credit scores, minorities being disadvantaged in security, or people not being recognized immediately by sensor technology, then it is not good (tech).?
This is exactly why we need to talk about Ethics and AI.?
The basic question in ethics theory is:?"What is the right thing to do?"
When we talk about?ethics in the context of?AI, we want to define how AI can do the right thing and live up to the values of our society. It is more than just a philosophical question, it is a socio-technological challenge and a business imperative to be solved. But why is it necessary to do that??
The details:?
The good news is that these ethical challenges have been recognized and are being addressed by many players in the AI field. In the following, we will look at the relevant aspects of developing ethical AI.?
1.?Fairness: The development and use of AI systems must be equitable, non-discriminatory, and ethical.
2. Responsibility: In AI systems, there must be the possibility to ensure and clearly assign responsibility and liability.
领英推荐
3. Benefit for society: AI systems must be used for the benefit of society while respecting societal values and human rights.
4. Data privacy: AI systems must respect users' privacy and data rights.?
5. Interdisciplinary cooperation and collaboration: The ethical aspects of AI must be researched and shaped jointly and inter-disciplinarily.
6. Transparency: AI systems must enable transparency for users.
7. Technical Robustness: AI systems must be designed and implemented to be technically robust and secure.
8. Sustainable human-machine cooperation: AI systems must be deployed in a way that promotes sustainable human-machine cooperation.
While some major IT players are struggling mightily with the intricacies of AI, others have focused on challenging potentially critical use cases and establishing methods, processes, and tools for trustworthy AI development. Examples include governance models such as AI Ethics Boards, tools such as AIEthics360 or OpenScale, methods such as Enterprise Design Thinking for AI, or the publication of principles and guides.?
We have the chance to shape AI and its applications. Now is the time to determine the future of AI and thus our society.
We can only do this by engaging in discourse - with all players in the field of AI.?This concerns politics, society, tech, users, scientists, industry, and we want to help start this discourse.?
From now on, we will discuss one of the above-mentioned guidelines every week as part of a campaign and look forward to your opinions, your input, and your participation in the discussion.?
--
Georg Olowson & Sophia Greulich
Advocates for Trustworthy AI, Consultants and Optimists
Note: This post and the views shared are the personal views of the authors.?
Great initiative you 2 ???????? happy to read more. What I usually get asked in the context: how can we (technically) identify bias and neutralize it? Is creating synthetic data to balance underrepresentation a good choice? Which methods can we use to understand the decision making of a CNN? Etc. These questions are a good sign, that companies are already wanting to “solve the problem” and not just talk about it ??. Let’s help them do so ??
Important Initiative. To tackle all these challenges we need interdisciplinary cooperation between researchers, industry, education, policy makers, and the civil society.
AI Governance Leader at IBM → Follow me to unlock trusted AI for your business | AI Partnerships | Speaker & Creator | Passionate Dancer
3 年Has been great to write this article with you - really insightful findings from your research! In the next weeks, we are going to go deeper into each of the topics! Naagma, Eric, Veronika
Focus on the customer
3 年Great initiative. The topic is so important, as more and more use cases for AI emerge. And totally agree, it needs to be an open discussion with everybody involved (not only technicians, but also other fields, maybe philosophers).