Following a short discussion in another forum, in which reference was made to AI having a Theory of Mind (ToM), I am interested in other opinions from RG members who have an active interest in an Ethics of AI.
In a sense, one of the critical emergent capabilities (in my thinking: limitations) is the idea that AI may possess a ToM. This presents an argument that any AI with a ToM exhibits a form of imagination, since it can attribute mental states to human subjects. However, the very act of such attribution is, to my mind, problematic in the least.
I have written, elsewhere, that “…through subjectification, [a] ToM assumes a rationality of action that may be irrationally violated by an ‘Other’. We are shaken when an absurd action is taken by an ‘Other’ that appears to us as irrational, or wrong, or immoral, or illegal; we question mental states we have ascribed to that ‘Other’, and whether they are an ‘other’ at all. Thus …the axiomatic variability of an individual’s mental state ensures there can be no level of universal access to reality…” Allowing AI to (imaginatively) attribute mental states as part of its outputs raises ethical concerns we perhaps have not yet begun to grasp. Is the real problem here that, perhaps ironically, generative AI is more human-like in its processing (flaws and all) than we might have anticipated.
Interested in other perspectives...
It'll be fine if you can link to the discussion. As a practitioner in this field and as a Mathematician,
AI having a Theory of Mind does not imply AI having the full functionality of the human mind. The "Imagination" concept can be ended here.
The human mind is a biological system and follows the basic principles of biological systems. One of them is :
Any biological system is highly complex and non-linear, can't be predicted and at most, it is possible to understand from where it'll become unpredictable. Biological systems may have infinite states and they can interact with each other. The Human mind is probably the most complex biological system having unknown numbers of neurons (states) interacting with each other.
AI is just a Mathematical Model (possibly) of a neural network and according to the definition of Mathematical Models, Mathematical Models can never predict the exact system outcome but provide an assumption about the outcome.
Simply, the definitions of Biological Systems, Mathematical Models and AI, we can say that, AI can never Imagine.
Another theory is, that AI is an Autonomous system and the human brain is non-autonomous. Imagination needs to illustrate some sort of scenario outside of the stored information in the memory/neurons.
As an Autonomous system, AI can not simply understand anything outside of its memory so AI Can Not Imagine.
Dear David Atkinson ,
Artificial intelligence (AI) is a wide-ranging tool that enables people to rethink how we integrate information, analyze data, and use the resulting insights to improve decision making—and already it is transforming every walk of life. In this report, Darrell West and John Allen discuss AI’s application across a variety of sectors, address issues in its development, and offer recommendations for getting the most out of AI while still protecting important human values.
https://www.brookings.edu/articles/how-artificial-intelligence-is-transforming-the-world/
Regards,
Shafagat
Arindam Kumar Paul, thank you for you contribution. I note your position from mathematical/positivistic science perspective, and there is much in your argument, however I am engaged more in a philosophical thought-experiment. I choose my words carefully to suggest a "form of imagination". This does not suggest that it is the same form of imagination a human possesses. Mathematics does not hold all the answers to all questions. While we wait for new theories to explain emergent phenomenon, thought experiments do have a place.
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Artificial Intelligence