The social implications of AI

Artificial Intelligence already influences several aspects of our society. Take a look at consumption habits, election results, interpersonal relations, trade, cities, culture, law, the military, the climate and more — in each field it is possible to see effects of AI either directly or indirectly.

It’s no secret, advanced technology plays a continually greater role in our lives and work. In this puzzle, some pieces could be explored further than has been the case until now, particularly the connection between AI, applied machine learning, other algorithms and social science. So how should we bridge this, and consider the larger effects on society as a whole by viewing these topics from an interdisciplinary perspective?

As of May 2018 there are 7.6 billion people in the world, and in the fourth quarter of the same year, Facebook had 2.3 billion monthly active users. Lately, the 2019 Hootsuite/WeAreSocial digital report came out, telling us that 57% of the world’s population now has internet access as 360 million new people went online for the first time. Something between 1.5–2.5 trillion searches happen on Google yearly, and the world’s digital community is predicted to spend 1.2 billion years in total using the internet in 2019. That means roughly 35–50% of the global population are being influenced by applied machine learning and narrow AI algorithms every month, while simultaneously getting value from usage.

These are quite abstract numbers due to their size, while also being concrete enough to show how larger parts of the global population has access to technology. But this access does in no way necessitate gaining the benefit of these technologies. Especially since they are used for gathering quantitative data used for influence and decision-making often regarding resources or large transnational interests that affect those who use these services. In other words: usage is contrasted by being used.

The mission of narrow AI

“When reading the news, you might see the terms ‘general’ and “narrow” AI. So what do these mean? Narrow AI refers to AI that handles one task. General AI, or Artificial General Intelligence (AGI) refers to a machine that can handle any intellectual task. All the AI methods we use today fall under narrow AI, with general AI being in the realm of science fiction.”
- Elements of AI (www.elementsofai.com)

There is something admirable about organizing the world’s information and making it universally accessible and useful (Google); to give people the power to build community and bring the world closer together (Facebook); or to be earth’s most customer-centric company; to build a place where people can come to find and discover anything they might want to buy online (Amazon). There are many more examples, all of which does provide some form of utility and at times agency.

Thus having more information than ever, being more connected and having easier access to most things sounds great when we say it out loud. Yet if we zoom in on different locations this has not necessarily been the case everywhere. Consumer insights are offered to companies by Google and Facebook or used actively by others such as Amazon. This insight has to be mined, a new form of extraction requiring both physical storage and vast amounts of energy. The best minds of our time are spending their life trying to market products to customers in a better way. Echo chambers of ‘likes’ begetting new targeted ‘likes’ contributing to polarisation; an algorithm can be categorised to a certain extent as racist in its judgement leading to imprisonment; and connections that show social comparison leading to mental health issues. The lofty ideas we had about technology as neutral have passed for some. Following this is an understanding that technology is not objective, not at all.

Technology is social

As such, if we now know that technology is social, a need arises for creating teams that consist of a diverse range of thinkers when working on these issues. Simply mixing in social scientists is additionally not enough, being educated in ethics does in no way guarantee ethical behaviour. Organisations today should work actively in small units across and between disciplines to facilitate a greater understanding of the various aspects that needs to be explored when we implement new technology.

As authors of this article, being engaged and interested, we are just students or recent graduates. Amidst climate change, weaponized artificial intelligence, refugee crises and more rising inequality — when most things seem surreal and truly difficult– it is precisely at this time when we need to gather and think. Think, so that the actions that we take are better, if only slightly better, and do our best to shift the course that we seem to be drifting down.

This is just the start, and there are some possible goals we can consider.

1. One goal may be to increase the collaboration between social thinkers and programmers in research efforts (As suggested by OpenAI).
2. A second goal could be to encourage a deeper understanding of programming for social thinkers and social science for programmers.
3. A third possible goal is creating platforms to gather this type of combined, interdisciplinary research and communicate it to students as well as businesses.
4. A fourth potential goal is taking part in development of solutions to monitor algorithms for possible behaviour that may disadvantage a population or prove dangerous for society. There are likely other possibilities for groups or goals.

Goal zero

Then there is poverty, water, urbanity, deprivation, etc. A list of issues we are facing locally that are now inevitably directly influenced by this technological push. If we were to forget these then all the aforementioned ambitions would be largely irrelevant. Is artificial intelligence a smokescreen in this sense? We tend to rapidly ascend to cloud-conversations of data or deep learning when the effects of these applications have some consequences we need to deeply consider. Technology in itself is not irresponsible. However, in the ‘agile’ environment of speed we find ourselves in, it is easy to forget why we were developing these novel solutions in the first place. It is challenging to propagate slow or ethical programming in this swift environment.

The goal zero could be to make AI work for our interplanetary shared ecology. Because social sciences, humanities, or computer science must not forget the world in which it is situated and depend on.

One incremental change on a platform with 100,000, one million or many billion users can slightly change how all of them act in real life. The indirect effects on society from the change of production processes in industry, lead to questions like: How will AI help optimize processes so that they use less of our natural resources? Can such optimization end up with us overusing another resource in the process? How will it affect our banking systems, our shopping habits, our energy system or our food production? All of these affect society and nature.

There are no policy makers or business leaders to influence first. It starts with you. If we can think of solutions or build platforms that is where it starts. It starts with collaboration between a few people. “Never doubt that a small group of thoughtful, committed citizens can change the world; indeed, it’s the only thing that ever has.” Margaret Mead said this — It is simple, obvious, and rings true today. Although we are influenced, and in the context of a rather large intangible ‘global’ society, it is a few people coming together in a variety of ways that facilitate the local steps staggering forward. We have a responsibility to do everything we can to understand our notions of progress while moving ahead.

AI Social Research is a small contribution, with a purpose of connecting thinkers together to engage in conversation and expression; to provide broader perspectives on AI and its effects on society through researching as an interdisciplinary group from backgrounds in social sciences, fin-tech, engineering, product development, renewable energy, psychology and informatics.