Here you can ask your questions about Artificial Intelligence. Below are the questions that have been received and answered by TRR 318 researchers so far.
Answers
“AI can certainly speed up a number of tasks, especially in areas such as publishing. However, where new content is created or decisions are made that affect people, AI should certainly not replace humans, i.e. control should remain with humans. Rather, AI is often best suited to support or prepare human activities, such as creating of draft texts or framework texts that are then reviewed, fleshed out and brought to life by humans. This can lead to a change in work processes and, in particular, optimization processes. AI makes different types of errors than humans, which can also affect copyright and data protection, for example. Humans will have to develop a new intuition for these errors, which can be more work, especially at the beginning. In the end, however, the benefits of properly applied AI will certainly often pay off in many ways.”
“AI technologies such as Midjourney and ChatGPT can now generate images and texts based on instructions that are almost indistinguishable from human work. In the case of images, it can be expected that AI will be increasingly used, especially where it is primarily used for support. Among other things, this means that accurate illustrations can be created and copyright issues can often be avoided. In the case of texts, for example, the creating recurring document types, summarizing of content or adapting its style are activities where AI can save a lot of time. In both cases, however, human intervention is still necessary when very specific (e.g. daily updated) content needs to be displayed or a certain quality needs to be guaranteed.
One danger that arises with AI in this context is its misuse for manipulation and misinformation, as the generated images and texts can hardly be distinguished from real ones. As a society, we are still ill-prepared for this; we need to relearn how to assess the trustworthiness of information.”
“AI can increasingly partially or fully automate activities that we previously assumed required human skills or intelligence. Activities that are essentially repetitive in nature can in many cases be done with AI, even if the specific content or information is new. Automation will enable us to take on many tasks that were previously often too time-consuming (e.g. giving individual feedback in education) and find solutions to problems that previously seemed too difficult (e.g. early detection and prevention of cancer). However, automation is expected to be accompanied by an intensification of work. Activities for which we used to take a lot of time are now done in the blink of an eye; as a result, our daily work will probably involve significantly more activities in the future, which can be exhausting.
AI also holds great opportunities as well as challenges for our society. AI can help us find solutions to climate change and previously incurable diseases and, in principle, it can involve everyone. However, its misuse in particular poses serious risks, such as individualized manipulation and misinformation in social media, mass surveillance and autonomous weapons systems. Politics, the economy and society must learn and practice a responsible approach to AI so that it ultimately benefits everyone. Hopefully, this can be achieved through the efforts of society as a whole.”
“The question is very rich in premises. The first part of the question concerns ‘working yourself to the bone’, which is of course very subjective. Objectively, the average working week for German employees has fallen by eight hours from 42.8 hours in 1965 to 34.7 hours in 2022. A key driver of the reduction in working hours is technological progress, which leads to higher productivity per hour. It is interesting to note that the subjective workload and perceived stress have remained almost constant over time. The reasons for this are certainly multifactorial; technostress could be one cause.
The second part of the question is whether artificial intelligence has the solution. This is also not so easy to say, because AI can of course take over certain tasks, but rarely complete jobs. At the same time, AI creates work for employees, because the technology has to be created and monitored, and inputs and outputs have to be monitored. It is therefore unlikely that human labor will become superfluous, but rather that job profiles will change. If AI is used well, it may be able to further reduce the necessary weekly working hours, but it remains subjective whether the work will then also ‘less working yourself to the bone’.”
“AI is a radical change in society with profound effects on how sociality will be understood and practiced. However, this is not the first time something like this happened. In the past, our society processed the profound impacts of other innovation, like written language, allowing us to communicate without having a human being in front of us. With written language we communicate with people far away from us, being unknown to us or even anonymous. Like nowadays, in these times people feared this will lead to the end of sociality. On the contrary, experience has shown the development of more articulate and complex forms of sociality.”
- Prof. Dr. Elena Esposito, project leder B01
“The controversial German philosopher Martin Heidegger once said about technology: where the danger (technology) lies, the saving power grows also. The mainstream view of AI is what can be called the ‘instrumental view’: AI is just a tool, and it can help us achieve human goals, such as curing new diseases. But AI, like other technologies, is a two-faced phenomenon, both a poison and a cure. As a poison, AI may threaten the social fabric. It has the capacity to infect the public discourse with fake news and misinformation, it may be weaponised by bad actors to undermine democratic processes, and it may divert people’s attention away from pressing social and political problems.
So is AI the beginning of the end of the social fabric? The unsatisfactory answer is ‘it depends.’ AI is not unique in its impact on the social fabric, so it is not really constitute a ‘beginning.’ Other media like the radio, television, and more recently the Internet, have profoundly shaken the social fabric and people are still trying to come to terms with its effects. For this reason, some schools are for instance trying to ban the use of cell phones and social media use. We can, however, cope with these changes, just as we can cope with the changes coming from AI. To do so, we need to investigate and respond to ethical issues like the ability of AI systems to treat people in discriminatory ways. Crucial for doing so is that we better understand and are able to explain how AI works and how it affects decision-making, together with citizens and other stakeholders. This is what the TRR aims to achieve, and it is in this sense, hopefully, the saving power will grow in the face of danger.”
- Dr. Wessel Reijers, Postdoctoral Researcher, B06
“The world will become human in a different way. The contribution of humans is essential for AI, even if our communication partner will not always be a human. Still, humans produce most of the material, used for training, regulating and testing algorithms. The challenge is to find new forms of control and participation.”
- Prof. Dr. Elena Esposito, project leader B01
“No. Thats why we need sophisticated ways of explaining, allowing us to check, evaluate and critizise AI-powered processes. This is the main motivation for our project.”
- Prof. Dr. Elena Esposito, project leader B01
“Certainly. It is the same way we forgot how to memorize long compositions like epic poems since we have writing available. However, we have not lost the ability to remember things, we just implement it in a different way. This is based on the availability of texts and classification systems, allowing us to access content without having to save it in our memory. Digital natives are not able to perform tasks that went without saying in the past, like reading maps. However, they have not lost the ability to achieve the same goal. The challenge is to combine de-skilling and re-skilling.”
- Prof. Dr. Elena Esposito, project leader B01
Questions in progress
In gaming, especially in strategy games, NPC enemies are completely stupid the whole time. On higher difficulty, they start cheating instead of performing better. Is it possible that progress in AI will lead to NPC-enemies that perform smarter tactics instead of cheating on higher difficulty?