Information professionals and intelligent machines – Can we still have librarians?
Artificial intelligence (AI) and the information age are bringing us more knowledge about ourselves and each other than any society has ever known. Yet at the same time it brings machines seemingly more capable of every human endeavour than any human can be. What are the limits of AI? Of intelligence and humanity more broadly? What are our ethical obligations to machines? Do these alter our obligations to each other? What is the basis of our social obligations? In this talk I will argue that there are really only two problems humanity (or any other species) has to solve. These are sustainability and inequality, or put another way, security and power. Or put a third way, how big of a pie can we make, and how do we slice up that pie. Life is not a zero-sum game; we and many other species use the security of sociality to construct public goods where everyone benefits. But still, every individual needs enough pie to thrive, and this is the challenge of inequality. I will argue that understanding these processes is not only essential to surviving the challenges of the climate crisis, but also helps answer the fundamental questions of ethics and social obligation. I will also examine how AI is presently affecting both of these problems. I will close with concrete policy recommendations for managing AI and our society.
Datasets make algorithms: how creating, curating, and distributing data creates modern AI
Recent advances in artificial intelligence have yielded powerful new generative models for text, speech, images, and video. These models have diverse applications in many fields. Yet, their power is derived from the data they are trained on. In fact, the creation of new AI models is constrained by the availability of properly curated, domain specific datasets. In this talk, I will show some recent examples of these models, and show how the same model can be used in different ways, when trained on different datasets. In some ways, the datasets are now part of the algorithms themselves. I will show how the work of creating, curating, and distributing datasets has direct implications for the advancement of AI. I will also cover some ethical issues involved in dataset creation and curation and their impacts on AI algorithms. I believe libraries can serve an expanded role in the creation and development of AI, and that building stronger working relationships between libraries and AI researchers will serve to benefit everyone.