The Five Major Areas of AI Ethics Before Superintelligence
There might be a reason to limit the media attention given to super AI, singularity and existential risk.
Many ethical concerns and problems related to artificial intelligence are still true, concrete and worth of discussing and researching. When studying and reviewing AI's concrete ethical problems it makes sense not just focus on the AI or machine learning itself, but link their influences to the long-lasting and ongoing era of digitalisation. AI and machine learning will continue the change and revolution which is started by internet, social media, smart phones and data-related solutions.
1. Distribution of income and employment
We believe that the change and challenge AI sets for our employment and distribution of income is huge. The tasks and jobs will can be automated and replaced is really high and under our current economical paradigm, this possibility will be utilized by our companies and public sector. Critics of this view usually refers to the luddities and are promising that a lot of new jobs will be available. Right now, that happening seems unlikely - the amount of the new task are really limited and the learning and know-how curve seems to be hard to handle for a big amount of our population.
We don't approach this change necessarily as a negative change. Our view is that there will be a need for a basic income solution and even a small level capital income will become more imporant. We see that the development in Finland is moving to this direction already.
When the share of work in human life decreases, we might walk into a problem of meaning. We have a long tradition of building our lives and societies around work life, but now we might have to change our mindset. Unpleasant news on the rise of the problems in mental health and drug use might tell us that some of us are already struggling with the new order in our societies. Some silent signals in the literature tells about the search of some kind of new spirituality in life. That might be something we need and we are looking for.
2. Data centralization and privacy
Corporate world has gone mad about data. Businesses are developming more and more data-centered. Hopefully the common practice will develop to the direction where the need for collecting data will be evaluated critically separately in every situation and solution.
Data seems to be centralizing to the hands of few major companies. IT and internet megacorporations possess an awful lot of information and data from us. People don't like this phenomenon, but they haven't learnt to oppose this development either.
Still, the data centralization remains only as a unpleasant risk, at least in the Western countries. The development taking in place in China puts us to think the more dangerous scenarios. It seems likely that we don't want that kind of the world where we will be seen only as large datasets and where our actions and behaviour is largely predictable and modifiable.
3. Security and safety
The development and implementation of AI create threats to our safety and security, too. These threats include small, practical safety and security issues, but also global concerns on our safety in a larger meaning. We have to learn to understand better safety and security matters related to the technology, starting from the everyday forms of cyber security.
Concerns on fully autonomous weapons and other AI warfare shape many questions on the global level. Outsourcing wars to robots might sound fascinating, but these kinds of AI solutions and agents contain dangerous and complex risks. A good thing is that the problems of AI warfare have been already recognized broadly by the multinational organisations and NGOs.
4. Algorithm and data biases
Algorithm and data biases can cause discrimination and unfair decisions. The transparency of algorithmic decision-making is an important area of ethical AI. Algorithms and data can easily contain biases whose origin is in human decision-making and these biases are difficult to detect, perceive and fix.
Algorithms and datasets are used for decion-making in recruiting, loan and mortgage applications and legal processes. All humans should be treated equally, but the historical data or the built algorithm can contain biases, which discrimate some sectors of the population.
Our understanding on algorithm and data biases is growing and developing. The topic is largely under research and for example, the problem had taken into account at the last NIPS conference.
5. Effects on human life
Even the applications of narrow AI will affect radically to our lives and how we see our world. The major discussion is how much power we want to give to technology and how much we let technology to direct our society and societal development. We still understand the digitalisation of the society very little.
The borderline between critical perspective on technological development and full-on technology optimism is fickle. It is about how we want to understand human life and value of humankind. The development of the last decades has made our world a lot more efficient, but equally irrefutably it has changed and eliminated some parts of our lives, especially forms of social intercourse.
When you follow the development of AI and robotics, it seems clear that social, care and sex robots and AIs capable of advanced conversation are reality very soon. It is likely that we should get used to see and meet robots even in our normal lives. How would you react if meeting a robot in the street?
Social robots will fulfil a lot of our needs in which we experience shortcomings. But does that isolate people even more from each other? Is it right, does it matter?