Ever dreamed of building your personalized AI to get really useful things done—and all without writing a single line of code? We explore the burgeoning realm of AI agents, showcasing how these tools are not just reshaping industries but revolutionizing our very approach to problem-solving and productivity. Drawing on extensive research and practical applications, the talk focuses on the real-world application of agentic AI to automate tasks, adapt decisions, and solve complex problems across various domains—from personal projects to professional endeavors. The journey with AI agents starts with foundational models like ChatGPT, Gemini, and Claude, which we personalize, contextualize with real-world tools, and then orchestrate into entire teams of AI agents. These multi-agent systems can tackle the hard problems that need to be resolved for reliable workflows, including hallucinations and robustness. The result mirrors the complexity of managing a company and a team of diverse AI agents. This leads to ethical considerations and the societal impacts of AI deployment, addressing valid concerns and common misconceptions, ranging from the fear of job displacement to threats to democracy. AI agents are not merely tools but partners in crafting our future. This keynote will empower you to harness these agents, enhancing your productivity and redefining your professional and personal landscapes. Whether you are a seasoned AI enthusiast or a newcomer to this field, you will leave with a profound understanding of how to navigate and shape the era of AI agents. 

The 70-year-long AI winter erupted into a burgeoning spring when, in 2023, ChatGPT became the fastest-spreading invention in history. Not only have generative AIs passed the historic Turing test of intelligence, but they are also taking our digital reality by storm, adding a potent ingredient to a cauldron of innovation already spiced up with blockchain and the 3D metaverse. The complexity of these modern AIs is staggering, and the detected productivity gains are enormous (over 50% for many intellectual tasks), while their emotional sensitivity surpasses that of average humans. Generative AIs are the last manifestation of the machine learning paradigm, which has automated knowledge creation at the click of a button. This development once again throws Homo sapiens into an identity crisis, as our species has traditionally defined itself by its previously held monopoly on the knowledge process. As knowledge automation finds its way into various sectors, from education to politics, this talk will dive into emerging opportunities and challenges. How can you position yourself to be part of this revolution? What could go wrong?

All of the world's most valuable companies call themselves "artificial intelligence" companies, exclusively. The role of algorithmic knowledge is not a vision of the future, but has been the dominant economic reality for years. The economic value that these companies create is to create 'digital twins' of economic processes and entire industries. Often, they take over existing industries by algorithmizing contained knowledge. Yes, it's still valuable to sell retail products, drive a taxi, rent a hotel bed, and write a song, but the added value consists of the knowledge created by Amazon, Uber, AirBnB, and Spotify in their digital twins of those industries. Its digital process models provide the playing field to pursue process optimization, resulting in more efficient, effective, and safer processes. Added value is created by maintaining a digital platform that contains knowledge of economic processes. In this workshop we begin with a review of the common features of the knowledge economy paradigm and then ask about the steps needed to implement this paradigm in the mining sector. We discussed the ubiquitous architectural feature of a "data star" at the information level and the "master algorithm" at the knowledge level. Empirical machine learning and theoretical computer simulations are the two main tools for optimizing knowledge about real-world processes in the field. We discuss the practical implications for a company's digital transformation.

The global pandemic has accelerated the current trend towards greater digitalization. Building on our established information and communication infrastructure, we managed our schools, commerce and online businesses during global lockdowns. As a next step beyond the digitization of information, many companies algorithmized knowledge processes, which goes a step further, from digital information to automated knowledge. We'll review these recent trends in the business world and discuss the power of automated knowledge platforms, based on machine learning. This forces us to review the limitations of this paradigm. There are systemic risks involved in over-reliance on artificial intelligence, not only ethical in nature, but also for economic gain. The knowledge economy requires a fusion of human intelligence and machine intelligence, of human vision and machine execution. Unfortunately, during this current phase, we often confuse means with ends. This, naturally, leads us to discuss the power that our artificial solutions already exert over us humans. Misinformation and fake news, addiction and mental health issues, political polarization, and commercial manipulation are all consequences of a growing imbalance of control between the artificial and the human. In the end, these technologies are extensions of the human mind. Maintaining control over global information and knowledge structures will also require the human mind to evolve to a new level. We discuss the implications.

With digital communication becoming our primary connection during times of social distancing, many of us are starting to feel our toxic relationship with algorithms. We've had numerous intellectual discussions on the subject over the past few years (e.g. here and here), but now it suddenly becomes personal. Can we take advantage of the current situation to develop a healthier relationship with our digital algorithms? The analogy between the pandemic and our relationship with our digital algorithms is apt. One of the few things that spread faster around the world than the virus is digital messages. Like COVID-19, our digital ecosystem poses a collective action problem. It can be a vector for spreading the virus, in the same way that it can be a vector for contagious fear and dangerous misinformation. Social algorithms bind our psychological health together. An exit strategy will require each individual to begin taking personal responsibility in a collective action of "distancing from social media." It does not mean stopping algorithmic interactions, but disinfecting them. As with a necessary trip to the grocery store, prepare appropriately the next time you get in close contact with your mobile phone. Just as the lack of an easy exit strategy for the pandemic requires the combination of several solutions, a more sustainable relationship with our algorithms will require a myriad of complementary measures, top-down and bottom-up. There are four possible alternatives. They are complementary and require everyone to take responsibility.

With more than 98% of the world's population using digital technology, human interaction produces an impressive digital footprint. This has rapidly transformed social and economic studies into true sciences, making predictions with 80-90% accuracy. Once data-poor, human science now offers the most comprehensive empirical evidence, covering nearly its entire sample continuously. While this creates unprecedented opportunities for businesses and policy, using digital fingerprinting for public policy faces challenges. The naive view of 'big data' imagines that computational skills alone would reveal reality in real time. In practice, data science resembles blind researchers touching different parts of an elephant, trying to piece together disparate evidence. Beyond computing challenges, data science integrates computer science, statistics, and application domains. Issues of representativeness, generalization, harmonization, variable definition, and data quality dominate practical data science. The talk concludes by examining the goal of creating data-driven flows for evidence-based policy decisions. Is automating the workflow from evidence to socio-economic interventions truly beneficial? If not, what else is needed?

The data science paradigm suggests that the future of decision-making consists of a fully automated workflow from observation, over computation, to a better world. This overlooks one of the most important limitations of empirical science: the fact that observational data always come from the past and can only inform the future if they are similar to the past (the technical concept is called "stationarity"). However, few of us aspire to a future similar to the past, even if it is an optimized version of the past. Today, machine learning is creating many of our most persistent injustices in its decision-making apparatus, including discrimination and inequality. Empirical data alone cannot help us break free from our established patterns. On the contrary, it locks us in them more and more firmly. In this talk, we explore this inherent limitation of empirical data science and talk about systematic ways to create the future we want.

Algorithms have begun to know us better than our best friends, partners, better than our mother, and better than ourselves. They can predict our personality, feelings, political opinion, sexual preferences, whether we use drugs, or whether our parents separated (or will separate?). Unlike human experts and geniuses, we can open up and look inside the "brains" of artificial intelligence and see what they do, when they do, what they do. We learn amazing things about human nature, about the most private core of ourselves, and about large-scale social processes. This talk explores some of the things that algorithms have taught us about ourselves over the past few years. 

Digital technology has not only revolutionized society but also the way we can understand it. An increasing share of human interaction leaves behind a massive digital footprint. Studying it allows us to gain unprecedented insights into what society is and how it works, including its intricate social networks that had long been obscure. Artificial intelligence allows us to detect hidden patterns through analytical tools such as machine learning and natural language processing. In addition, computer simulations allow us to explore and explain hypothetical situations that may not even exist in reality, but that we would like to exist: a better world. Traditionally, if a social study could explain about 10-20% of the variance in a phenomenon, it was published in the most prestigious journals, and policies were implemented that implemented these findings. Many of them failed. Over the past few years, we have begun to predict human and social behavior with 80-90% accuracy. Social studies is becoming a science. What are the consequences?

Human beings compute information, biologically and artificially, and so do societies. The growing fusion of social and digital networks, of analog and digital footprints, of biological and artificial intelligence, shows us how many tasks that were once in the intangible domain of implicit human behavior are easily encoded by explicit algorithms. If social entities process information, and if an amazing part of it can be absorbed by digital technology, then the fundamental theories behind those technologies could also help us understand how society communicates and computates information. I use basic concepts from information theory and theoretical computer science and apply them to different aspects of social behavior. We can measure the amount of information that is communicated from humans to algorithms and vice versa. Moreover, this perspective illustrates that society has always calculated, even long before the digital age. Laws, routines, cultural habits, prejudices, and psychological biases can be conceptualized as algorithms and information channels. This perspective makes it easier to understand how we can research, guide, and shape the often intimidating and overwhelming developments of the digital revolution.