We present an overview of higher randomness and its recent developments. After an introduction, we provide in the second section some background on higher computability, presenting in particular $\Pi^1_1$ and $\Sigma^1_1$ sets from the viewpoint of the computability theorist. In the third section we give an overview of the different higher randomness classes: $\Delta^1_1$-randomness, $\Pi^1_1$-Martin-Löf randomness, higher weak-2 randomness, higher difference randomness, and $\Pi^1_1$-randomness. We then move on to study each of these classes, separating them and inspecting their respective lowness classes. We put more attention on $\Pi^1_1$-Martin-Löf randomness and $\Pi^1_1$-randomness: The former is the higher analogue of the most well-known and studied class in classical algorithmic randomness. We show in particular how to lift the main classical randomness theorems to the higher settings by putting continuity in higher reductions and relativisations. The latter presents, as we will see, many remarkable properties and does not have any analogue in classical randomness. Finally in the eighth section we study randomness along with a higher hierarchy of complexity of sets, motivated by the notion of higher weak-2 randomness. We show that this hierarchy collapses eventually.

In this introductory survey, we provide an overview of the major developments of algorithmic randomness with an eye towards the historical development of the discipline. First we give a brief introduction to computability theory and the underlying mathematical concepts that later appear in the survey. Next we selectively cover four broad periods in which the primary developments in algorithmic randomness occurred: (1) the mid-1960s to mid-1970s, in which the main definitions of algorithmic randomness were laid out and the basic properties of random sequences were established; (2) the 1980s through the 1990s, which featured intermittent and important work from a handful of researchers; (3) the 2000s, during which there was an explosion of results as the discipline matured into a fully-fledged subbranch of computability theory; and (4) the early 2010s, in which ties between algorithmic randomness and other subfields of mathematics were discovered. The aim of this survey is to provide a point of entry for newcomers to the field and a useful reference for practitioners.

The halting probability of a Turing machine was introduced by Chaitin, who also proved that it is an algorithmically random real number and named it Omega. Since his seminal work, many popular expositions have appeared, mainly focusing on the metamathematical or philosophical significance of this number (or debating against it). At the same time, a rich mathematical theory exploring the properties of Chaitin's Omega has been brewing in various technical papers, which quietly reveals the significance of this number to many aspects of contemporary algorithmic information theory. The purpose of this survey is to expose these developments and tell a story about Omega which outlines its multi-faceted mathematical properties and roles in algorithmic randomness.

This is a survey of constructive and computable measure theory with an emphasis on the close connections with algorithmic randomness. We give a brief history of constructive measure theory from Brouwer to the present, emphasizing how Schnorr randomness is the randomness notion implicit in the work of Brouwer, Bishop, Demuth, and others. We survey a number of recent results showing that classical almost everywhere convergence theorems can be used to characterize many of the common randomness notions including Schnorr randomness, computable randomness, and Martin-Löf randomness. Last, we go into more detail about computable measure theory, showing how all the major approaches are basically equivalent (even though the definitions can vary greatly).

In this survey, we lay out the central results in the study of algorithmic randomness with respect to biased probability measures. The first part of the survey covers biased randomness with respect to computable measures. The central technique in this area is the transformation of random sequences via certain randomness-preserving Turing functionals, which can be used to induce non-uniform probability measures. The second part of the survey covers biased randomness with respect to non-computable measures, with an emphasis on the work of Reimann and Slaman on the topic, as well as the contributions of Miller and Day in developing Levin's notion of a neutral measure. We also discuss blind randomness as well as van Lambalgen's theorem for both computable and non-computable measures. As there is no currently-available source covering all of these topics, this survey fills a notable gap in the algorithmic randomness literature.

Ergodic theory is concerned with dynamical systems -- collections of points together with a rule governing how the system changes over time. Much of the theory is concerned with the long term behavior of typical points-- how points behave over time, ignoring anomalous behavior from a small number of exceptional points. Computability theory has a family of precise notions of randomness: a point is "algorithmically random'' if no computable test can demonstrate that it is not random. These notions capture something essential about the informal notion of randomness: algorithmically random points are precisely the ones that have typical orbits in computable dynamical systems. For computable dynamical systems with or without assumptions of ergodicity, the measure 0 set of exceptional points for various theorems (such as Poincaré's Recurrence Theorem or the pointwise ergodic theorem) are precisely the Schnorr or Martin-Löf random points identified in algorithmic randomness.

We discuss the different contexts in which relativization occurs in randomness and the effect that the relativization chosen has on the results we can obtain. We study several characterizations of the K-trivials in terms of concepts ranging from cuppability to density, and we consider a uniform relativization for randomness that gives us more natural results for computable randomness, Schnorr randomness, and Kurtz randomness than the classical relativization does (the relativization for Martin-Löf randomness is unaffected by this change). We then evaluate the relativizations we have considered and suggest some avenues for further work.

Algorithmic randomness lies at the intersection between computability theory and probability theory. In order to fully explore this interaction, one naturally needs a computable version of measurable functions. While several such notions appear in the literature, most of them do not interact well with algorithmic randomness because they are only defined up to a null set. Therefore, we need a computable notion of measurable function which is well defined on algorithmically random points, and this is what layerwise computability precisely does. This article is a survey about this notion. We give the main definitions, the most important properties, and several applications of this notion. We prioritize motivating this framework and explaining its salient features.

The field of algorithmic randomness studies the various ways to define the inherent randomness of individual points in a space (for example, the Cantor space or Euclidean space). Classically, this quest seems quixotic. However, the theory of computing allows us to give mathematically meaningful notions of random points. In the past few decades, algorithmic randomness has been extended to make use of resource-bounded (e.g., time or space) computation. In this survey we survey these developments as well as their applications to other parts of mathematics.

Algorithms are often disparaged, but they are also sometimes overrated. To understand their place in our world it is also important to understand their limits.

In the age of algorithms, new fears have arisen, including the fear that one day we will be overtaken, or even enslaved, by new beings, transhumans favored by natural selection, but also, more simply, by computers or algorithms more intelligent than us. This leads us to a question that computer scientists have been asking since the 1950s: Can an algorithm be intelligent? This question brings up two others. What does the adjective intelligent mean? Can we create an intelligent being?