News on the limits of AI and alternatives

My university published an interview about my views on the limits of AI and what I think are better alternatives for technological development.

Here is a short video clip:

Roundtable talk on the problem of meaning in AI

I was invited to give a presentation on the problem of meaning in artificial intelligence as part of an international roundtable on machine learning, artificial intelligence, and super-computation.

Here is the official poster with the details:

The Problem of Meaning in AI and Robotics: Still with Us after All These Years

Fittingly published in the 10-year anniversary of the publication of “enactive AI“, here is a critical retrospective piece that at the same time marks a significant departure into new, largely unexplored directions. Exciting times!

The Problem of Meaning in AI and Robotics: Still with Us after All These Years

Tom Froese and Shigeru Taguchi

In this essay we critically evaluate the progress that has been made in solving the problem of meaning in artificial intelligence (AI) and robotics. We remain skeptical about solutions based on deep neural networks and cognitive robotics, which in our opinion do not fundamentally address the problem. We agree with the enactive approach to cognitive science that things appear as intrinsically meaningful for living beings because of their precarious existence as adaptive autopoietic individuals. But this approach inherits the problem of failing to account for how meaning as such could make a difference for an agent’s behavior. In a nutshell, if life and mind are identified with physically deterministic phenomena, then there is no conceptual room for meaning to play a role in its own right. We argue that this impotence of meaning can be addressed by revising the concept of nature such that the macroscopic scale of the living can be characterized by physical indeterminacy. We consider the implications of this revision of the mind-body relationship for synthetic approaches.

Invited talk: The problem of meaning in AI and robotics

I was invited to give a talk at the conference cycle of the Cognitive Robotics Laboratory. The conference will celebrate the lab’s 10th anniversary, and will take place Feb. 21-22 at UAEM in Cuernavaca. Here is my title and abstract:

The problem of meaning in AI and robotics

Tom Froese

In recent years there has been a lot of renewed excitement about the possibilities of creating advanced artificial intelligence (AI) that could rival the human mind. I cast doubt on this prospect by reviewing past revolutions in cognitive robotics, specifically the shift toward embodied cognition in the 90s and the recent emphasis on the enactive approach. I argue that despite claims to the contrary, these revolutions did not manage to overcome the fundamental problem of meaning, which was first identified following the various theoretical and practical problems faced by Good Old-Fashioned AI. Similarly, even after billions of dollars of investment, today’s commercial computational systems simply do not understand anything in the way that humans or, so I argue, even the simplest living creatures do. I therefore propose a paradigm shift in how to conceptualize the overall vision and goals of the synthetic method: we should stop aiming to replicate human understanding with AI, and instead focus on helping humans better realize their potential via human-computer interfaces, including robotic systems.

Review of “Contemporary Sensorimotor Theory”

ContemporarySensorimotorTheoryAs part of the Frontiers in Robotics and AI research topic “Re-enacting sensorimotor experience for cognition” we published a book review on this topic.

Book review: Contemporary sensorimotor theory

Tom Froese and Franklenin Sierra

Consciousness, with its irreducible subjective character, was almost exclusively a philosophical topic until relatively recently. Today, however, the problem of explaining the felt quality of experience has also become relevant to science and engineering, including robotics and AI: “What would we have to build into a robot so that it really felt the touch of a finger, the redness of red, or the hurt of a pain?” (O’Regan, 2014, p. 23). Yet a practical response still requires an adequate theory of consciousness, which brings us back to the hard problem: how can we account, from a scientific point of view, for the phenomenological character of experience? Over a decade ago, O’Regan and Noë (2001) proposed a new approach to these questions, the so-called sensorimotor approach to perceptual experience. How far has this approach come and what are its outstanding challenges? The volume Contemporary Sensorimotor Theory, edited by Bishop and Martin, takes stock of the current state of the field.