11 Comments
User's avatar
Manu Sharma's avatar

Brilliant! Extremely insightful article, and a timely reminder (thankfully constant on this Substack by Dr. Vishnoi) that we stand to lose a lot - potentially everything, if we actually lose what it means to be human - if we let ourselves be seduced by the power of what AI can help us see and mistake it for something it is not.

Expand full comment
Naresh Vishnoi's avatar

Very elaborate and distinctive roles of human mind/intellect as different from AI.

Expand full comment
Rajesh Gopakumar's avatar

Great post! A footnote physics comment (which only strengthens the points you are making): One of Maxwell's great achievements, in some sense going beyond the ones you mentioned, is the introduction of the field as an abstract but fundamental physical entity with its own dynamics untethered to any material object. It liberated physics from the newtonian paradigm of particle motion being everything. It was also an important feat of abstraction to talk of the local field since most of the earlier known laws of electromagnetism (Gauss, Ampere) were formulated as integrals - in terms of currents and electric/magnetic fluxes.

Expand full comment
Vijay Chandru's avatar

“We are prone to overestimate how much we understand about the world.”

- Daniel Kahneman

Thank you Nisheeth for this wonderful thought provoking post. Took me back 30 years to conversations with my mentor Professor Roddam Narasimha. The renowned fluid dynamicist and historian of Indian classical science who called what we seem to celebrate these days “computational positivism”

Would urge you and others interested in these questions to read

Narasimha, R. (2003). The Indian half of Needham’s question: some thoughts on axioms, models, algorithms, and computational positivism. Interdisciplinary Science Reviews, 28(1), 54–66. https://doi.org/10.1179/030801803225010340

It may be time to think about a synthesis of how thinking slow (or reframing as you call it) can be combined with thinking fast may lead to a true 5th paradigm of science.

Expand full comment
Nisheeth Vishnoi's avatar

Thank you for this thoughtful response and for the pointer to Roddam Narasimha’s work, which I hadn't encountered. “Computational positivism” is such an apt phrase, and I look forward to reading the essay.

The connection you make between Kahneman’s thinking slow and conceptual reframing is compelling, and the kind of synthesis I hope more of us can explore together.

Expand full comment
Maxim Raginsky's avatar

Vijay, Nisheeth, you both might like my essay about computational positivism, philosophy of science, and their relations to AI:

https://realizable.substack.com/p/on-computational-positivism

Expand full comment
Nisheeth Vishnoi's avatar

Thanks for the pointer Maxim, will take a look!

Expand full comment
Gangan Prathap's avatar

What is this 5th Paradigm?

Expand full comment
Nisheeth Vishnoi's avatar

Thank you! That’s a wonderful addition, and beautifully put. The move to treat the field as a fundamental entity was a conceptual leap, and a striking example of abstraction reshaping the foundations of physics. Thanks again for deepening the essay's arc!

Expand full comment
Gangan Prathap's avatar

Observation -> Patterns -> Empirical Laws and Principles -> Theoretical Laws. These are the two loops from Observation to Discovery. When the two laws converge, we have closure. That is my experience. Both loops need intuitive leaps, comprising induction, deduction, and abduction (for want of a better term). F = ma was discovered as an empirical law. The derivation from Least Action Principles was equally difficult and that effort ended with Noether linking invariance and symmetry to conservation.

Expand full comment
Rakesh Biswas's avatar

To summarise drastically what appears to (me ) to be the problem statement (and one proposed solution in the comments) here is:

Reframing or slow processing or system 2 in Kahneman's dual process framework, appears to be a human thing at present but there's a possibility of perhaps designing it into the current cognitive framework of machines?

Expand full comment