The problem of meaning in AI: Still with us after all these years

I was invited to give a talk at the “Programs, minds and machines” workshop, which will be hosted jointly by the Mathematics and the Philosophy Research Institutes of UNAM, August 6-9, 2018.

The problem of meaning in AI: Still with us after all these years

Tom Froese

In recent years there has been a lot of excitement about the possibilities of advanced artificial intelligence that could rival the human mind. I cast doubt on this prospect by reviewing past revolutions in cognitive robotics, specifically the shift toward situated robotics in the 90s and the shift toward a dynamical approach in the 00s. I argue that despite claims to the contrary, these revolutions did not manage to overcome the fundamental problem of meaning that was first identified in the context of various theoretical and practical problems faced by Good Old-Fashioned AI. Even after billions of dollars of investment, today’s computers simply do not understand anything. I argue for a paradigm shift in the field: the aim should not be to replicate the human mind in autonomous systems, but to help it realize its full potential via interfaces.

Advertisements

An artificial life approach to the origins of the genetic code

I have been invited to give a talk at the “Special workshop: The Earth, Life and Artificial Life”, sponsored by ELSI, which will take place next Friday, July 27, as part of the International Conference on Artificial Life in Tokyo.

The title and abstract are as follows:

An artificial life approach to the origins of the genetic code

Tom Froese

A growing number of artificial life researchers propose that making progress on the problem of the origins of life requires taking seriously life’s embodiment: even very simple life-like systems that are spatially individuated can interact with their environment in an adaptive manner. This behavior-based approach has also opened up new perspectives on a related unsolved problem, namely the origin of the genetic code, which can now be seen as emerging out of iterated interactions in a community of individuals. Thus, artificial life demonstrates that the dominant scientific strategy of searching for the conditions of Darwinian evolution should be broadened to consider other possibilities of optimization.

Is there room for intrinsic normativity in a synthetic system?

I was invited to give a keynote talk at the workshop “The Synthetic Approach to Biology and the Cognitive Sciences (SA-BCS 2018): Developing an Epistemology for the Synthetic Sciences of Life and Cognition“, which will take place as part of ALIFE 2018 in Tokyo on July 25.

Here are the title and abstract of my contribution:

Is there room for intrinsic normativity in a synthetic system?

Tom Froese

Enactivism rejects the standard hypothesis of cognitive science, according to which all cognition involves the unconscious manipulation of mental representations, and instead replaces it with a dynamical systems account. And yet enactivism also resists purely dynamical approaches that see no role for any kind of subjectivity, because it appeals to the role of our lived phenomenology and claims that living beings behave with respect to intrinsic norms directed at maintaining their self-produced viability. So far, this middle way seems to be philosophically unsatisfactory: at best it allows us to claim that acting in accordance with experience or norms just is identical to a certain kind of dynamic pattern. But this turns subjectivity into a mysterious difference that makes no difference with respect to the unfolding of those patterns, which remain completely determined by the dynamical laws alone. This calls for deeper epistemological reflection about how it could be possible for subjectivity to play a role in an objective world, while avoiding a regression to the untenable positions of either representationalism or eliminativism. This debate has implications for the synthetic method, especially regarding longstanding discussions about the relative merits of software, hardware, and wetware.

Talk on water processions and pulque inebriations

A Teotihuacan Mini-Symposium will take place in San Juan Teotihuacan next Monday, July 2. I am going to present the latest advances of my modeling attempts regarding the ancient city’s hypothetical collective sociopolitical network. Here are the title and abstract:

Water processions and pulque inebriations: Simulating the effects of communal ritual on social coordination

Tom Froese

In previous work I simulated a sociopolitical network of the hypothesized collective government of early Teotihuacan based on the distribution of Three-Temple Complexes. Given that actors make decisions and change relationships with others primarily in self-interested ways, it was found that network-wide coordination of decisions is nearly impossible to achieve. However, it was also demonstrated that this social coordination problem is consistently overcome when actors occasionally participate in communal rituals, particularly of the chaotic form in which people’s normal social constraints are temporarily bracketed and replaced by spontaneous, and often intoxicated, individualized behaviors. These fiesta-like rituals allowed network-wide social coordination to arise in a self-organized manner. Yet a type of ritual much more prominently depicted in murals, processions, involves exactly the opposite: highly organized and conventionalized movements that are executed by several individuals in synchrony. In this talk I will compare the extent to which these types of ritual could have facilitated social coordination. The results suggest that processions may have been less effective as long as the sociopolitical network consisted of highly modular clusters of actors, which suggests that they only became an important ritual form after Teotihuacan became a more integrated city.

New paper on iterated learning at the origins of life

Jorge, Nathaniel and I have published an extension of our iterated learning approach to the origins of the genetic code in the Proceedings of the Artificial Life Conference 2018. We unexpectedly found that the most likely sequences in which amino acids get incorporated into the emerging genetic codes in our simulation model exhibit a remarkable overlap with the sequence predicted in the literature based on empirical considerations.

We will present this work at the ALIFE conference in Tokyo as part of the special session on “Hybrid Life: Approaches to integrate biological, artificial and cognitive systems”.

An iterated learning approach to the origins of the standard genetic code can help to explain its sequence of amino acid assignments

Tom Froese, Jorge I. Campos, and Nathaniel Virgo

Artificial life has been developing a behavior-based perspective on the origins of life, which emphasizes the adaptive potential of agent-environment interaction even at that initial stage. So far this perspective has been closely aligned to metabolism-first theories, while most researchers who study life’s origins tend to assign an essential role to RNA. An outstanding challenge is to show that a behavior-based perspective can also address open questions related to the genetic system. Accordingly, we have recently applied this perspective to one of science’s most fascinating mysteries: the origins of the standard genetic code. We modeled horizontal transfer of cellular components in a population of protocells using an iterated learning approach and found that it can account for the emergence of several key properties of the standard code. Here we further investigated the diachronic emergence of artificial codes and discovered that the model’s most frequent sequence of amino acid assignments overlaps significantly with the predictions in the literature. Our explorations of the factors that favor early incorporation into an emerging artificial code revealed two aspects: an amino acid’s relative probability of horizontal transfer, and its relative ease of discriminability in chemical space.

Figure 2

Illustration of the architecture of the genetic system of one of our hypothetical protocells.

Psychological study on chaos control

Dobri Dotov and I have published an extended abstract in the Proceedings of the Artificial Life Conference 2018 about the study that he realized at UNAM. We suggest that the results have implications for how we should think about how to stabilize the behavior of complex adaptive systems with which we can interact.

We will present this work at the ALIFE conference in Tokyo as part of the special session on “ALife and Society: Transcending the artificial-natural divide”.

Mutual synchronization and control between artificial chaotic system and human

Dobromir Dotov and Tom Froese

Dexterous assistive devices constitute one of the frontiers for hybrid human-machine systems. Manipulating unstable systems requires task-specific anticipatory dynamics. Learning this dynamics is more difficult when tasks, such as carrying liquid or riding a horse, produce unpredictable, irregular patterns of feedback and have hidden dimensions not projected as sensory feedback. We addressed the issue of coordination with complex systems producing irregular behaviour, with the assumption that mutual coordination allows for non-periodic processes to synchronize and in doing so to become regular. Chaos control gives formal expression to this: chaos can be stabilized onto periodic trajectories provided that the structure of the driving input takes into account the causal structure of the controlled system.

Can we learn chaos control in a sensorimotor task? Three groups practiced an auditory-motor synchronization task by matching their continuously sonified hand movements to sonified tutors: a sinusoid served as a Non-Interactive Predictable tutor (NIP), a chaotic system stood for a Non-Interactive Unpredictable tutor (NI-U), and the same system weakly driven by the participant’s movement stood for an Interactive Unpredictable tutor (I-U). We found that synchronization, dynamic similarity, and causal interaction increased with practice in I-U. Our findings have implications for current efforts to find more adequate ways of controlling complex adaptive systems.

UNISON

Keynote at “Time, the Body, and the Other”

I will be a keynote speaker at the international conference on Time, the Body, and the Other: Phenomenological and Psychopathological Approaches, which will take place September 13-15, 2018, in Heidelberg.

The title of my contribution will be: “Integrating Phenomenology and Systems Theory: Time and the Other in Schizophrenia as a Case Study”

Header-September-2018neu

Talk at the Annual Meeting of the Society for American Archaeology

Next week I will present the latest installment of our model-based research into the social organization of ancient Teotihuacan at this year’s Annual Meeting of the Society for American Archaeology. Here is the title and abstract:

A network model of co-rulership and community ritual in Teotihuacan: From neighborhoods to districts

Tom Froese and Linda R. Manzanilla

Experts remain divided about the nature of the sociopolitical system of ancient Teotihuacan, which was one of the earliest and largest urban civilizations of the Americas. Excavations hoping to find compelling evidence of a powerful dynasty of rulers, such as a royal tomb, keep coming away empty-handed. However, the alternative possibility of a corporate or collective government, perhaps headed by a small number of co-rulers, also remains poorly understood. A third option is that the city’s collective government begun as a fully decentralized network of neighborhood representatives, but this kind of arrangement seems susceptible to the problems of cooperation and action coordination. Previously we used a computational model to show that in principle this latter worry is unfounded, as long as we assume that the network’s topology could be transformed via community rituals and was not strongly subdivided (Froese, Gershenson, and Manzanilla 2014). Here we extend this model to investigate whether centralized hierarchy could mitigate the negative effects of strong divisions. The new results reveal a peculiar synergy between hierarchy and community ritual in that only their combination improved the extent of coordination, which is consistent with portrayals of the elite as religious specialists serving the public.

Keynote at From Animals to Animats 15 (SAB 2018)

I will be a keynote speaker at FROM ANIMALS TO ANIMATS 15: The 15th International Conference on the Simulation of Adaptive Behavior (SAB 2018), which will take place 14-17 August 2018, in Frankfurt, Germany, and is organized by the International Society for Adaptive Behavior (ISAB).

Here is my title and abstract:

Searching for the conditions of genuine intersubjectivity: From robotics to HCI

Tom Froese

Many our most valued experiences are experiences that we share with others. Yet the basis for this sense of we-ness remains mysterious. Could it really be possible that two people share one and the same experience? How so? Two lines of research are providing important insights. First, complex systems analyses of social robotics and agent-based models have demonstrated that there is nothing mysterious about the possibility of cognitive activity being distributed in a multi-agent system. Second, experimental investigations of real-time embodied social interaction mediated by human-computer interfaces demonstrate that co-regulation of interaction dynamics makes a difference to experience. This formal and empirical research on social interaction supports the possibility of genuine intersubjectivity: we can directly participate in the unfolding of each other’s experience.

The 2nd Week on Complexity Sciences at C3-UNAM

The 2nd Week on Complexity Sciences will be held at the Center for Complexity Sciences (C3) at UNAM’s main campus from Jan. 31 to Feb 2. There will be many international invited speakers.

I will give a talk on the recent work I did with Prof. Alejandro Frank on the origins of the genetic code on Jan. 31 at 13:00. The title of our contribution is “A new approach to the origin of the genetic code”.

« Older entries