Kristinn R. Thórisson

Reykjavík University · Department of Computer Science

The development of artificial intelligence (AI) systems has to date been largely one of man-ual labor. This constructionist approach to AI has resulted in systems with limited-domain application and severe performance brittleness. No AI architecture to date incorporates, in a single system, the many features that make natural intelligence general-p...

The concept of understanding is commonly used in everyday communications , and seems to lie at the heart of human intelligence. However, no concrete theory of understanding has been fielded as of yet in artificial intelligence (AI), and references on this subject are far from abundant in the research literature. We contend that the ability of an ar...

Research into the capability of recursive self-improvement typically only considers pairs of \(\langle \)agent, self-modification candidate\(\rangle \), and asks whether the agent can determine/prove if the self-modification is beneficial and safe. But this leaves out the much more important question of how to come up with a potential self-modifica...

Dependable cyber-physical systems strive to deliver anticipative, multi-objective performance anytime, facing deluges of inputs with varying and limited resources. This is even more challenging for life-long learning rational agents as they also have to contend with the varying and growing know-how accumulated from experience. These issues are of c...

Many existing AGI architectures are based on the assumption of infi-nite computational resources, as researchers ignore the fact that real-world tasks have time limits, and managing these is a key part of the role of intelligence. In the domain of intelligent systems the management of system resources is typi-cally called "attention". Attention mec...

Research on general machine intelligence is concerned with building machines that are capable of performing a multitude of highly complex tasks in environments as complex as the real world. A system placed in such a world of indefinite possibilities and never-ending novelty must be able to adjust its plans dynamically to adapt to changes in the env...

Causal knowledge and reasoning allow cognitive agents to predict the outcome of their actions and infer the likely reasons behind observed events, enabling them to interact with their surroundings effectively. Causality has been the subject of some research in artificial intelligence (AI) over the past decade due to its potential for task-independe...

Significant increases in industry requirements for network bandwidth are seen year after year. The exponential growth in streaming data is matched by an increase in the use of machine learning and deep learning to glean actionable insights from these data—ideally in real-time. Demand for artificial intelligence (AI) solutions to a variety of comput...

Traditional computer vision models often require extensive manual effort for data acquisition and validation, particularly when detecting subtle behavioral nuances or events. The difficulty in distinguishing routine behaviors from potential risks in real-world applications, like differentiating routine shopping from potential shoplifting, further c...

Humanity is currently facing one of its biggest challenges to date: The climate crisis. As a result, most industry sectors are reassessing their ways of working to be better equipped to address their share of the situation. The digital sector often gets set aside in such considerations in talk about the green transition because a significant amount...

Explanation can form the basis, in any lawfully behaving environment , of plans, summaries, justifications, analysis and predictions, and serve as a method for probing their validity. For systems with general intelligence, an equally important reason to generate explanations is for directing cumulative knowledge acquisition: Lest they be born knowi...

A system deployed in the real world will need to handle uncertainty in its observations and interventions. For this, we present an approach to introduce uncertainty of state variables in causal reasoning using a constructivist AI architecture. Open questions of how noisy data can be handled and intervention uncertainty can be represented in a causa...

A system deployed in the real world will need to handle uncertainty in its observations and interventions. For this, we present an approach to introduce uncertainty of state variables in causal reasoning using a constructivist AI architecture. Open questions of how noisy data can be handled and intervention uncertainty can be represented in a causa...

Explanation can form the basis, in any lawfully behaving environment, of plans, summaries, justifications, analysis and predictions, and serve as a method for probing their validity. For systems with general intelligence, an equally important reason to generate explanations is for directing cumulative knowledge acquisition: Lest they be born knowin...

Humanity is currently facing one of its biggest challenges to date: The climate crisis. As a result, most industry sectors are reassessing their ways of working to be better equipped to address their share of the situation. The digital sector often gets set aside in such considerations in talk about the green transition because a significant amount...

Auditing is a field of expertise often mentioned as being ripe for automation using artificial intelligence methods at all levels of operations. Primarily, the application of artificial intelligence (AI) in the auditing profession is done by and for large organizations, leveraging large datasets. While AI approaches for big data are continually imp...

The papers in this collection were presented at the third annual International Workshop on Self-Supervised Learning (IWSSL-22), held at Reykjavik Uni- versity in Reykjavik, Iceland, on July 28th and 29,th 2022. Weighing in together at a total of 127 pages, these 10 papers of exceptional quality describe leading AI research from Europe and the USA,...

Making analogies is a kind of reasoning where two or more things are compared, to highlight or uncover attributes of interest. Besides being useful for comparing what is known, analogy making can help a learning agent deal with tasks and environments not experienced before, where similarities and differences to known phenomena and their cause-effec...

What sets artificial intelligence (AI) apart from other fields of science and technology is not what it has achieved so far, but rather what it set out to do from the very beginning, namely, to create autonomous self-contained systems that can rival human cognition-machines with 'human-level general intelligence.' To achieve this aim calls for a ne...

A cognitive architecture aimed at cumulative learning must provide the necessary information and control structures to allow agents to learn incrementally and autonomously from their experience. This involves managing an agent's goals as well as continuously relating sensory information to these in its perception-cognition information processing st...

Tasks are of primary importance for artificial intelligence (AI), yet no theory about their characteristics exists. The kind of task theory we envision is one that allows an objective comparison of tasks, based on measurable physical properties, and that can serve as a foundation for studying, evaluating, and comparing learning controllers of vario...

Without a concrete measure of the “complicatedness” of tasks that artificial agents can reliably perform, assessing progress in AI is difficult. Only by providing evidence of progress towards more complicated tasks can developers aiming for general machine intelligence (GMI) ascertain their progress towards that goal. No such measure for this exist...

Any machine targeted for human-level intelligence must be able to autonomously use its prior experience in novel situations, unforeseen by its designers. Such knowledge transfer capabilities are usually investigated under an assumption that a learner receives training in a source task and is subsequently tested on another similar target task. Howev...

A cognitive architecture aimed at cumulative learning must provide the necessary information and control structures to allow agents to learn incrementally and autonomously from their experience. This involves managing an agent's goals as well as continuously relating sensory information to these in its perception-cognition information stack. The mo...

This paper introduces a new metamodel-based knowledge representation that significantly improves autonomous learning and adaptation. While interest in hybrid machine learning / symbolic AI systems leveraging, for example, reasoning and knowledge graphs, is gaining popularity, we find there remains a need for both a clear definition of knowledge and...

This paper introduces a new metamodel-based knowledge representation that significantly improves autonomous learning and adaptation. While interest in hybrid machine learning / symbolic AI systems leveraging, for example, reasoning and knowledge graphs, is gaining popularity, we find there remains a need for both a clear definition of knowledge and...

The knowledge that a natural learner creates based on its experience of any new situation is likely to be both partial and incorrect. To improve such knowledge with increased experience , cognitive processes must bring already-acquired knowledge towards making sense of new situations and update it with new evidence, cumulatively. For the initial cr...

Autonomous knowledge transfer from a known task to a new one requires discovering task similarities and knowledge generalization without the help of a designer or teacher. How transfer mechanisms in such learning may work is still an open question. Transfer of knowledge makes most sense for learners for whom novelty is regular (other things being e...

While several tools exist for training and evaluating narrow machine learning (ML) algorithms, their design generally does not follow a particular or explicit evaluation methodology or theory. Inversely so for more general learners, where many evaluation methodologies and frameworks have been suggested, but few specific tools exist. In this paper w...

The complex socio-technological debate underlying safety-critical and ethically relevant issues pertaining to AI development and deployment extends across heterogeneous research subfields and involves in part conflicting positions. In this context, it seems expedient to generate a minimalistic joint transdisciplinary basis disambiguating the refere...

While several tools for training and evaluating narrow machine learning (ML) algorithms exist, their design generally does not follow a particular or explicit evaluation methodology or theory. Inversely so for more general learners, where many evaluation methodologies and frameworks have been suggested but few if any specific tools exist. In this p...

An important feature of human learning is the ability to continuously accept new information and unify it with existing knowledge, a process that proceeds largely automatically and without catastrophic side-effects. A generally intelligent machine (AGI) should be able to learn a wide range of tasks in a variety of environments. Knowledge acquisitio...

An important feature of human learning is the ability to continuously accept new information and unify it with existing knowledge, a process that proceeds largely automatically and without catastrophic side-effects. A generally intelligent machine (AGI) should be able to learn a wide range of tasks in a variety of environments. Knowledge acquisitio...

In the quest for artificial general intelligence (AGI), questions remain about what kinds of representations are needed for the kind of flexibility called for by complex environments like the physical world. A capacity for continued learning of many domains has yet to be realized, and proposals for how to achieve general performance improvement thr...

A generally intelligent machine (AGI) should be able to learn a wide range of tasks. Knowledge acquisition in complex and dynamic task-environments cannot happen all-at-once, and AGI-aspiring systems must thus be capable of cumulative learning: efficiently making use of existing knowledge during learning, supporting increases in the scope of abilit...

A generally intelligent machine (AGI) should be able to learn a wide range of tasks. Knowledge acquisition in complex and dynamic task-environments cannot happen all-at-once, and AGI-aspiring systems must thus be capable of cumulative learning: efficiently making use of existing knowledge during learning, supporting increases in the scope of abilit...

In the quest for artificial general intelligence (AGI), questions remain about what kinds of representations are needed for the kind of flexibility called for by complex environments like the physical world. A capacity for continued learning of many domains has yet to be realized, and proposals for how to achieve general performance improvement thr...

While many evaluation procedures have been proposed in past research for artificial general intelligence (AGI), few take the time to carefully list the (minimum, general) requirements that an AGI-aspiring (cognitive) control architecture is intended to eventually meet. Such requirements could guide the design process and help evaluate the potential...

We consider the role of reasoning in a resource-limited controller that explicitly and continuously models its environment, and uses these models as a basis for its prediction and action. Several important features of such cumulative modeling are identified, with an emphasis on how abduction and deduction can be used to continuously prune and refin...

We report on a series of new platforms and events dealing with AI evaluation that may change the way in which AI systems are compared and their progress is measured. The introduction of a more diverse and challenging set of tasks in these platforms can feed AI research in the years to come, shaping the notion of success and the directions of the fi...

Artificial intelligence (AI) and machine learning (ML) research has traditionally focused most energy on constructing systems that can learn from data and/or environment interactions. This paper considers the parallel science of teaching: Artificial Pedagogy (AP). Teaching provides us with a method—aside from programming—for imparting our knowledge...

The concept of “common sense” (“commonsense”) has had a visible role in the history of artificial intelligence (AI), primarily in the context of reasoning and what’s been referred to as “symbolic knowledge representation.” Much of the research on this topic has claimed to target general knowledge of the kind needed to ‘understand’ the world, storie...

While evaluation of specialized tools can be restricted to the task they were designed to perform, evaluation of more general abilities and adaptation requires testing across a large range of tasks. To be helpful in the development of general AI systems, tests should not just evaluate performance at a certain point in time, but also facilitate the...

The concept of “task” is at the core of artificial intelligence (AI): Tasks are used for training and evaluating AI systems, which are built in order to perform and automatize tasks we deem useful. In other fields of engineering theoretical foundations allow thorough evaluation of designs by methodical manipulation of well understood parameters wit...

The concept of understanding is commonly used in everyday communications , and seems to lie at the heart of human intelligence. However, no concrete theory of understanding has been fielded as of yet in artificial intelligence (AI), and references on this subject are far from abundant in the research literature. We contend that the ability of an ar...

The concept of “task” is at the core of artificial intelligence (AI): Tasks are used for training and evaluating AI systems, which are built in order to perform and automatize tasks we deem useful. In other fields of engineering theoretical foundations allow thorough evaluation of designs by methodical ma- nipulation of well understood parameters w...

Evaluation of artificial intelligence (AI) systems is a prerequisite for comparing them on the many dimensions they are intended to perform on. Design of task-environments for this purpose is often ad-hoc, focusing on some limited aspects of the systems under evaluation. Testing on a wide range of tasks and environments would better facilitate comp...

Out of fear that artificial general intelligence (AGI) might pose a future risk to human existence, some have suggested slowing or stopping AGI research, to allow time for theoretical work to guarantee its safety. Since an AGI system will necessarily be a complex closed-loop learning controller that lives and works in semi-stochastic environments,...

A data construct called a semcard is a semantic (meaning-based) software object including semantic meta-tags and meta-data that describes a target object or thing. A target object can be any type of digital or physical entity or identifier, or it can be tacit knowledge, such as ideas, concepts, processes or other data existing in a user's mind, pro...

Among other disclosure, a knowledge network and semcards enabling intelligent matching of offers and requests, involving all types of information and knowledge, including information such as classified ads, data about products and services, or knowledge, expertise, ideas, suggestions, opinions, and other forms of tacit knowledge is described.

A significant problem facing researchers in reinforcement learning, and particularly in multi-objective learning, is the dearth of good benchmarks. In this paper, we present a method and software tool enabling the creation of random problem instances, including multi-objective learning problems, with specific structural properties. This tool, calle...

We present an architectural approach to learning problem solving skills from demonstration, using internal models to represent problem-solving operational knowledge. Internal forward and inverse models are initially learned through active interaction with the environment, and then enhanced and finessed by observing expert teachers. While a single i...

Four principal features of autonomous control systems are left both unaddressed and unaddressable by present-day engineering methodologies: (1) The ability to operate effectively in environments that are only partially known at design time; (2) A level of generality that allows a system to reassess and redefine the fulfillment of its mission in lig...

Humans and other animals are often touted as examples of systems that possess general intelligence. However, rarely if ever do they achieve high levels of intelligence and autonomy on their own: they are raised by parents and caregivers in a society with peers and seniors, who serve as teachers and examples. Current methods for developing artificia...

An important part of human intelligence is the ability to use language. Humans learn how to use language in a society of language users, which is probably the most effective way to learn a language from the ground up. Principles that might allow an artificial agents to learn language this way are not known at present. Here we present a framework wh...

In the domain of intelligent systems the management of system resources is typically called “attention”. Attention mechanisms exist because even environments of moderate complexity are a source of vastly more information than available cognitive resources of any known intelligence can handle. Cognitive resource management has not been of much conce...

Four principal features of autonomous control systems are left both unaddressed and unaddressable by present-day engineering methodologies: 1. The ability to operate effectively in environments that are only partially known beforehand at design time;; 2. A level of generality that allows a system to re-assess and re-define the fulfillment of its mi...

In this paper we consider the issue of endowing an AGI system with decision-making capabilities for operation in real-world environments or those of comparable complexity. While action-selection is a critical function of any AGI system operating in the real-world, very few applicable theories or methodologies exist to support such functionality, wh...

Resource-boundedness must be carefully considered when designing and implementing artificial general intelligence (AGI) algorithms and architectures that have to deal with the real world. But not only must resources be represented explicitly throughout its design, an agent must also take into account their usage and the associated costs during reas...

Systems intended to operate in dynamic, complex environments --- without intervention from their designers or significant amounts of domain-dependent information provided at design time --- must be equipped with a sufficient level of existential autonomy. This feature of naturally intelligent systems has largely been missing from cognitive architec...

We present a cognitive architecture whose main constituents are allowed to grow through a situated experience in the world. Such an architectural growth is bootstrapped from a minimal initial knowledge and the architecture itself is built around the biologically-inspired notion of internal models. The key idea, supported by findings in cognitive ne...

A key goal in designing an artificial intelligence capable of performing complex tasks is a mechanism that allows it to efficiently choose appropriate and relevant actions in a variety of situations and contexts. Nowhere is this more obvious than in the case of building a general intelligence, where the contextual choice and application of actions...

The development of artificial intelligence (AI) systems has to date taken largely a constructionist approach, with manual programming playing a central role. After half a century of AI research, enormous gaps persist between artificial and natural intelligence. The differences in capabilities are readily apparent on virtually every scale we might w...

Much of present AI research is based on the assumption of computational systems with infinite resources, an assumption that is either explicitly stated or implicit in the work as researchers ignore the fact that most real-world tasks must be finished within certain time limits, and it is the role of intelligence to effectively deal with such limita...

One of the original goals of artificial intelligence (AI) research was to create machines with very general cognitive capabilities and a relatively high level of autonomy. It has taken the field longer than many had expected to achieve even a fraction of this goal; the community has focused on building specific, targeted cognitive processes in isol...

This chapter discusses hierarchically organised actions in communication. One essential, but often overlooked, feature of natural dialogue is turn taking. More recently, turn taking has become an issue in robot and virtual human research as researchers aim to make these systems more fluent and dynamic when interacting naturally with humans. The mos...

We present an architectural approach to learning problem solving skills from demonstration, using internal models to represent problem-solving operational knowledge. Internal forward and inverse models are initially learned through active interaction with the environment, and then enhanced and finessed by observing expert teachers. While a single i...

This book constitutes the refereed proceedings of the 11th International Conference on Intelligent Virtual Agents, IVA 2011, held in Reykjavik, Island, in September 2011. The 18 revised full papers and 27 revised short papers presented together with 25 poster papers were carefully reviewed and selected from 91 submissions. The papers are organized...

A number of present-day problems act to hold back progress in the field of artificial intelligence (A.I.), both theoretical and pragmatic. Among the most serious pragmatic issues has to do with integration and large-scale systems construction, as much recent work on humanoids and interactive robots has

Many dialogue systems have been built over the years that address some subset of the many complex factors that shape the behavior of participants in a face-to-face conversation. The Ymir Turntaking Model (YTTM) is a broad computational model of conversational skills that has been in development for over a decade, continuously growing in the number...

Multimodal natural behavior of humans presents a complex yet highly coordinated set of interacting processes. Providing robots with such interactive skills is a challenging and worthy goal and numerous such efforts are currently underway; evaluating the progress in this direction, however, continues to be a challenge. General methods for measuring...