Humans and AI systems are usually portrayed as separate sys- tems that we need to align in values and goals. However, there is a great deal of AI technology found in non-autonomous systems that are used as cognitive tools by humans. Under the extended mind thesis, the functional contributions of these tools become as essential to our cognition as our brains. But AI can take cognitive extension towards totally new capabil- ities, posing new philosophical, ethical and technical chal- lenges. To (...) analyse these challenges better, we define and place AI extenders in a continuum between fully-externalized systems, loosely coupled with humans, and fully-internalized processes, with operations ultimately performed by the brain, making the tool redundant. We dissect the landscape of cog- nitive capabilities that can foreseeably be extended by AI and examine their ethical implications. We suggest that cognitive extenders using AI be treated as distinct from other cognitive enhancers by all relevant stakeholders, including developers, policy makers, and human users. (shrink)

This paper attempts to answer a central question in unsupervised learning: what does it mean to “make sense” of a sensory sequence? In our formalization, making sense involves constructing a symbolic causal theory that both explains the sensory sequence and also satisfies a set of unity conditions. The unity conditions insist that the constituents of the causal theory – objects, properties, and laws – must be integrated into a coherent whole. On our account, making sense of sensory input is a (...) type of program synthesis, but it is unsupervised program synthesis. Our second contribution is a computer implementation, the Apperception Engine, that was designed to satisfy the above requirements. Our system is able to produce interpretable human-readable causal theories from very small amounts of data, because of the strong inductive bias provided by the unity conditions. A causal theory produced by our system is able to predict future sensor readings, as well as retrodict earlier readings, and impute (fill in the blanks of) missing sensory readings, in any combination. In fact, it is able to do all three tasks simultaneously. We tested the engine in a diverse variety of domains, including cellular automata, rhythms and simple nursery tunes, multi-modal binding problems, occlusion tasks, and sequence induction intelligence tests. In each domain, we test our engine's ability to predict future sensor values, retrodict earlier sensor values, and impute missing sensory data. The Apperception Engine performs well in all these domains, significantly out-performing neural net baselines. We note in particular that in the sequence induction intelligence tests, our system achieved human-level performance. This is notable because our system is not a bespoke system designed specifically to solve intelligence tests, but a general-purpose system that was designed to make sense of any sensory sequence. (shrink)

In the last 20 years the Turing test has been left further behind by new developments in artificial intelligence. At the same time, however, these developments have revived some key elements of the Turing test: imitation and adversarialness. On the one hand, many generative models, such as generative adversarial networks, build imitators under an adversarial setting that strongly resembles the Turing test. The term “Turing learning” has been used for this kind of setting. On the other hand, AI benchmarks are (...) suffering an adversarial situation too, with a ‘challenge-solve-and-replace’ evaluation dynamics whenever human performance is ‘imitated’. The particular AI community rushes to replace the old benchmark by a more challenging benchmark, one for which human performance would still be beyond AI. These two phenomena related to the Turing test are sufficiently distinctive, important and general for a detailed analysis. This is the main goal of this paper. After recognising the abyss that appears beyond superhuman performance, we build on Turing learning to identify two different evaluation schemas: Turing testing and adversarial testing. We revisit some of the key questions surrounding the Turing test, such as ‘understanding’, commonsense reasoning and extracting meaning from the world, and explore how the new testing paradigms should work to unmask the limitations of current and future AI. Finally, we discuss how behavioural similarity metrics could be used to create taxonomies for artificial and natural intelligence. Both testing schemas should complete a transition in which humans should give way to machines—not only as references to be imitated but also as judges—when pursuing and measuring machine intelligence. (shrink)

The extended mind thesis maintains that the functional contributions of tools and artefacts can become so essential for our cognition that they can be constitutive parts of our minds. In other words, our tools can be on a par with our brains: our minds and cognitive processes can literally ‘extend’ into the tools. Several extended mind theorists have argued that this ‘extended’ view of the mind offers unique insights into how we understand, assess, and treat certain cognitive conditions. In this (...) chapter we suggest that using AI extenders, i.e., tightly coupled cognitive extenders that are imbued with machine learning and other ‘artificially intelligent’ tools, presents both new ethical challenges and opportunities for mental health. We focus on several mental health conditions that can develop differently by the use of AI extenders for people with cognitive disorders and then discuss some of the related opportunities and challenges. (shrink)

The main factor of intelligence is defined as the ability tocomprehend, formalising this ability with the help of new constructsbased on descriptional complexity. The result is a comprehension test,or C- test, which is exclusively defined in computational terms. Due toits absolute and non-anthropomorphic character, it is equally applicableto both humans and non-humans. Moreover, it correlates with classicalpsychometric tests, thus establishing the first firm connection betweeninformation theoretical notions and traditional IQ tests. The TuringTest is compared with the C- test and the (...) combination of the two isquestioned. In consequence, the idea of using the Turing Test as apractical test of intelligence should be surpassed, and substituted bycomputational and factorial tests of different cognitive abilities, amuch more useful approach for artificial intelligence progress and formany other intriguing questions that present themselves beyond theTuring Test. (shrink)

We present nine facets for the analysis of the past and future evolution of AI. Each facet has also a set of edges that can summarise different trends and contours in AI. With them, we first conduct a quantitative analysis using the information from two decades of AAAI/IJCAI conferences and around 50 years of documents from AI topics, an official database from the AAAI, illustrated by several plots. We then perform a qualitative analysis using the facets and edges, locating AI (...) systems in the intelligence landscape and the discipline as a whole. This analytical framework provides a more structured and systematic way of looking at the shape and boundaries of AI. (shrink)

The Minimum Description Length principle is the modernformalisation of Occam's razor. It has been extensively and successfullyused in machine learning, especially for noisy and long sources ofdata. However, the MDL principle presents some paradoxes andinconveniences. After discussing all these, we address two of the mostrelevant: lack of explanation and lack of creativity. We present newalternatives to address these problems. The first one, intensionalcomplexity, avoids extensional parts in a description, so distributingcompression ratio in a more even way than the MDL principle. (...) The secondone, information gain, forces that the hypothesis is informative wrt. the evidence, so giving a formaldefinition of what is to discover. (shrink)

Animals, including humans, are usually judged on what they could become, rather than what they are. Many physical and cognitive abilities in the ‘animal kingdom’ are only acquired (to a given degree) when the subject reaches a certain stage of development, which can be accelerated or spoilt depending on how the environment, training or education is. The term ‘potential ability’ usually refers to how quick and likely the process of attaining the ability is. In principle, things should not be different (...) for the ‘machine kingdom’. While machines can be characterised by a set of cognitive abilities, and measuring them is already a big challenge, known as ‘universal psychometrics’, a more informative, and yet more challenging, goal would be to also determine the potential cognitive abilities of a machine. In this paper we investigate the notion of potential cognitive ability for machines, focussing especially on universality and intelligence. We consider several machine characterisations (non-interactive and interactive) and give definitions for each case, considering permanent and temporal potentials. From these definitions, we analyse the relation between some potential abilities, we bring out the dependency on the environment distribution and we suggest some ideas about how potential abilities can be measured. Finally, we also analyse the potential of environments at different levels and briefly discuss whether machines should be designed to be intelligent or potentially intelligent. (shrink)

Artificial Intelligence is making rapid and remarkable progress in the development of more sophisticated and powerful systems. However, the acknowledgement of several problems with modern machine learning approaches has prompted a shift in AI benchmarking away from task-oriented testing towards ability-oriented testing, in which AI systems are tested on their capacity to solve certain kinds of novel problems. The Animal-AI Environment is one such benchmark which aims to apply the ability-oriented testing used in comparative psychology to AI systems. Here, we (...) present the first direct human-AI comparison in the Animal-AI Environment, using children aged 6–10. We found that children of all ages were significantly better than a sample of 30 AIs across most of the tests we examined, as well as performing significantly better than the two top-scoring AIs, “ironbar” and “Trrrrr,” from the Animal-AI Olympics Competition 2019. While children and AIs performed similarly on basic navigational tasks, AIs performed significantly worse in more complex cognitive tests, including detour tasks, spatial elimination tasks, and object permanence tasks, indicating that AIs lack several cognitive abilities that children aged 6–10 possess. Both children and AIs performed poorly on tool-use tasks, suggesting that these tests are challenging for both biological and non-biological machines. (shrink)

This paper defines in a formal and computational way the notion of ‘consilience’, a term introduced by Whewell in 1847 for the evaluation of scientific theories. Informally, as has been used to date, a model or theory is ‘consilient’ if it is predictive, explanatory and unifies the evide-nce. Centred in a constructive framework, where new terms can be intro-duced, we essay a formalisation of the idea of unification based on the avoidance of ‘sepa-ration’. However, it is soon manifest that this (...) classical approach is vulnerable to the introduction of fantastic concepts to unify disparate sub-theories. Our second approach is constructed by using a detailed evaluation of the relationship between the theory and the evid-ence by means of reinforcement propagation. With the use of reinfor-cement, fantastic concepts can be better detected and the role of con-silience for theory construction and revision can be specialised for dif-ferent inference mechanisms like explanatory induction, abduction, deduc-tion and analogy. (shrink)

Mundus vult decipi, ergo decipiatur—the world wants to be deceived, so let it be deceived.Artificial intelligence has been a deceiving discipline: AI addresses those tasks that, if performed by humans, would require intelligence, but have been solved without featuring any genuine intelligence. This delusion has come, in return, with algorithmic techniques that can reliably solve many of these tasks, from game playing to pattern recognition. AI applications are a success.However, AI has not solved “what makes [us] clever, the puzzle of (...) intelligence”. Those that expect that this book brings the answer will have their hopes deceived. This book is about what does not make us clever, a restatement of Fodor’s “The mind doesn’t work that way” . Partridge swings between condescending scepticism and bitter negativism with most of what AI has become, and reaches fierce disparagement of the technological singularity visionaries.The main message of the book is that science is stil .. (shrink)