In AI, it is challenging to contextualize facts about the environment properly. An approach to the reasoning for modifying information and behaviour in response to novel circumstances is needed to build AI. 

When studying AI, the frame problem is the most critical obstacle to overcome. If the frame problem can be solved, it opens the door to building intelligent machines with agency.

Frame problem

In AI, the frame problem is challenging with using first-order logic (FOL) to transmit facts about a robot in the real world. FOL is when a sentence or statement predicate can only relate to one topic. It is also known as first-order functional calculus or first-order predicate calculus. Traditional FOL uses several hypotheses to represent a robot's state by showing that environmental components do not change randomly. The frame problem is the difficulty of assembling suitable collections of axioms to describe a robot environment.

Philosophical problems

John McCarthy and Patrick J. Hayes first articulated this issue in their 1969 article, Some Philosophical Problems from the Perspective of AI. The formal mathematical problem was used as a jumping-off point for broader explorations of the challenge of knowledge representation for AI in this study and others like it. 

Since computers are the managers of compartmentalized and predetermined data processing, Dreyfus and Dreyfus argue they need to understand what has changed and what has remained the same. The frame problem demonstrates the need to participate in thinking and reasoning actively. It necessitates a reliable database that can supply AI with readily available information.

Three aspects of the frame problem

The primary issue with the frame problem is determining the best way to express the link between a set of rules and behaviours. An exact definition of the frame problem cannot be found. We have yet to find a purpose all philosophers and AI researchers agreed upon. There are several definitions of the frame problem. This variation is due to differing perspectives on categorizing the frame problem. The frame problem can be divided into three categories: metaphysical, logical, and epistemological.

Metaphysical aspect

The Metaphysical Aspect of the Frame Problem concerns practical investigations to discover and implement general laws for ordinary world experience. The spatiotemporal features of environmental data should be included in this practical research. These practical studies show how to update views about the world when an agent encounters a novel event. Cognitive science, mainly extracting information from areas of an agent's cognitive activity, is regarded as a component of these practical studies. Pattern recognition, for example, can be viewed as a metaphysical element of the frame problem. Numerous descriptions can be used to provide a metaphysical account of the frame problem. For example, Janlert saw the frame problem as a metaphysical challenge in 1988 because it is concerned with the form and internal workings of the representation rather than the instrumental appropriateness of representation.

Conclusion

In the late 1980s and early 1990s, the narrow, technical frame problem inspired a considerable lot of work in logic-based AI, and its broader philosophical implications emerged about the same time. However, each thinker's emphasis on the frame problem today is usually determined by their position on other issues.

The logical frame problem has been solved in various ways within classical AI, and it is no longer regarded as a severe barrier, even for those working in a strictly logic-based paradigm. It's worth noting that logically-minded AI researchers can maintain their methodology while rejecting the traditional cognitive scientist's belief in the importance of computation over representations for understanding the mind. Furthermore, because the purpose of classical AI is not to create computers with human-level intelligence but rather to create better and more useful computer programmes, it is immune to the philosophical objections of Fodor, Dreyfus, and others. 

Significantly, for AI researchers operating outside of the paradigm of symbolic representation entirely—for example, those working on situated robotics—the logical frame problem does not appear in day-to-day inquiries.

Image source: Unsplash