Updates and News from the Center for Humans and Machines
Connect AI Berlin
Participants at the CHM's networking event Connect AI Berlin.
On November 17th, we hosted Connect AI Berlin at Harnack House, an event that brought together Berlin's leading researchers in machine and behavior research. The day was characterized by insightful discussions, building connections, and facilitating collaborative research and innovation in the dynamic landscape of AI research.
Between a compelling keynote delivered by Moritz Hardt from the Max Planck Institute for Intelligent Systems, thought-provoking presentations by Berlin-based scientists, scholars, and researchers, and an engaging workshop delving into the future and challenges of LLMs, the event immersed participants in a rich tapestry of insights and discussions at the crossroads of AI, society, and innovation. Grateful for the opportunity, we eagerly anticipate the continued impact of these connections on shaping discourse and driving positive transformations in Berlin's research landscape and beyond.
P.S. Here you can find some more pictures from the day!
Machine & Behavior Conference
Building on the momentum, we are excited to announce the upcoming Machine + Behavior Conference on April 29-30, 2024, also taking place at the iconic Harnack House. Following the success of Connect AI, this exclusive event aims to unite scientists from different domains to address pivotal challenges in a post-ChatGPT era.
Amidst the evolving landscape of AI, we will explore pressing issues such as behavioral AI safety, the future of work, and human communication. This is a call to action to collectively shape the research agenda on "Machine Behavior" in the wake of Generative AI's transformative impact, building on what we started with this Nature article.
Submissions for the poster track
If you wish to take part in this conference, we are currently accepting submissions for the poster track. This platform provides an opportunity for researchers, particularly those early in their careers, to showcase their latest results, ongoing projects, and innovative ideas in a succinct, high-impact manner. Selected posters will be prominently featured at the event, providing a unique platform for collaboration and exchange of ideas – here you can find more information along with the submission link.
Join us as we collaboratively chart research priorities, build a robust community in machine behavior research, and lay the groundwork for the future. We look forward to your participation in this exciting venture.
Welcome to the Humans and Machines – a newsletter all about interdisciplinary research that seeks to understand, anticipate, and shape major disruptions from digital media and artificial intelligence to how we think, learn, work, play, cooperate, and govern.
This Is Us
Some of the CHM members at the team retreat.
Highlights of the Summer
The summer in the Northern Hemisphere is over, but we’re still thinking back to all the exciting moments. Here are some of the most notable ones:
Looking back to IC2S2 2023
Levin Brinkmann giving a plenary talk.
In July, we had an incredible opportunity to attend the International Conference for Computational Social Science (IC2S2) 2023 in Copenhagen, a major event bringing together social and computational science experts to analyze and understand our social world using large-scale data and computational models.
The conference, held at the Mærsk Tower of the University of Copenhagen, was a welcomed opportunity for our researchers to connect and exchange ideas with leading experts from various fields, including economics, computer science, and psychology. The MPIB being the second-largest group attending the conference, our team left a mark with some thought-provoking presentations: Levin Brinkmann delivered "Exploring the Evolution of Artistic Styles Using Generative AI and Influence Modeling," Nils Kobis shared insights on the "Ethical Risks of Delegation to Artificial Intelligence," Tobias Werner presented an interesting take on "Algorithmic and Human Collusion," and Edmond Awad offered a deep-dive into "Mapping Charitable Giving Preferences from 3 Million Decisions Worldwide."
Our researchers also engaged in enriching discussions during poster sessions, offering attendees a glimpse into our latest work. Our IC2S2 experience was rewarding, and we look forward to applying these insights to our future research efforts.
The Horizon Europe Grant
Nils Köbis, © MPIB
Nils Kobis, a senior research scientist at the CHM, was recently awarded an EU Horizon Grant for the project RESPOND – Rescuing Democracy from Political Corruption in Digital Societies. Nils is part of an international consortium led by the University of Bologna that has secured a remarkable €6 million in funding.
The project focuses on leveraging digital technology to fight against corruption, aiming to produce new insights, explore the practices of political corruption in both national and transnational political systems, understand the impact of grey zones of corruption on citizens' perception and engagement in democracy and, of course, develop a set of tools that can tackle these problems and help curb political corruption. Nils will conduct research centering around the use of technology, such as AI, to combat corruption at both national and transnational levels, building upon previous collaborative work with Iyad Rahwan on the role of AI in fighting corruption.
The Horizon Europe program epitomizes European cooperation, emphasizing scientific and technical excellence while addressing pressing global challenges. This grant marks a significant milestone in ongoing efforts to uphold integrity and effectively combat political corruption.
Research Spotlight: What risks emerge out of the interaction between humans and intelligent machines, and how can they be mitigated?
In this newsletter, we’re delving into one of our current research areas – Behavioral Safety & Ethics.
In our rapidly evolving world, AI has become an integral part of our daily lives, from children's toys to home assistants like Amazon's Alexa, influencing over 100 million users. AI taking on responsibilities on behalf of humans, from setting prices in online markets to conducting interrogations, is also becoming increasingly common. Within this transformative landscape, we explore fundamental questions that shape the future of AI: Could machines go astray and corrupt human behavior? How can we design AI systems to minimize ethical and safety risks? And how do people around the world perceive these risks?
An illustration of the main roles through which intelligent actors can corrupt ethical behavior, along with the summary of the main fears and mechanisms of each role. The color coding indicates the strength of the corrupting force of AI: not reaching human levels yet (green), reaching but not surpassing human levels (yellow), and surpassing human levels (red).
If these questions intrigue you as they do us, we invite you to take a look at our ongoing research on our website. Join us on this journey as we seek to understand and navigate the ethical and safety dimensions of AI in an increasingly AI-driven world.
What we’re working on at the moment
Despite people's disregard for AI advice promoting honesty, they’re surprisingly willing to follow advice that encourages breaking ethical norms, even when they are aware of its source. The AI Advisor explores how individuals respond to AI advice encouraging rule-breaking. It appears that algorithmic advisors can significantly influence people's behavior in ethical dilemmas, leading to increased unethical behavior, similar to human advisors' impact. Moreover, the approach of algorithmic transparency does not prove effective in reducing this effect.
What AI might look like: The Face Game
Have you ever wondered how AI will select its face for interactions on social media?
The Face Game, a collaborative multidisciplinary research endeavor created and launched by researchers from the Max Planck Institute for Human Development, the Toulouse School of Economics, the University of Exeter, and the University of British Columbia, together with the Universidad Autonoma de Madrid and Université Paris Cité, delves into this intriguing question. This game explores how AI can choose and adapt its digital identity, depending on its goals and the individuals it interacts with, which has significant implications for the future of digital interactions with AI. However, your participation is crucial for achieving this goal, so play the game and take part in defining AI's role in our digital world.
What we’re inspired by
💫 As AI-powered conversational agents become more human-like, users increasingly view them as companions. In the study conducted by Pataranutaporn et al. (2023), individuals who perceived the AI they interacted with as caring saw it as as more trustworthy, empathetic, and effective, highlighting the he importance of how AI systems are introduced for shaping interactions and the overall experience with the AI.
🤖 While the generative AI chatbots are already capable of outperforming humans on various creative tasks, this study by Koivisto & Grassini (2023) found that the best human ideas still matched or even exceeded those of the chatbots. It emphasizes AI's capacity to boost innovative thinking while recognizing the distinctive complexity of human creative processes, informing essential questions about the future of imaginative work in the AI era.
🔮 According to Zahiu et al. (2023), it might be possible to leverage VR to train empathy and moral growth in ways that surpass the capabilities of conventional moral education. This approach offers the possibility to provide users with immersive, emotionally engaging experiences that mimic real-life scenarios without overriding their personal agency or capacity for reflection.