"AI Company Consults Religious Scholars on Machine Consciousness"
AI alignment
ensuring an AI system's goals and behaviour match human values and intentions.
Mechanistic interpretability
research into the internal workings of neural networks to explain model outputs.
Artificial intelligence research has reached a stage where companies train models capable of humanlike conversation, reasoning, and even autonomous digital action. This raises a central governance question called the alignment problem: ensuring an AI system's goals and behaviour match human values and intentions, rather than diverging from them in ways developers did not intend. Large language models are trained on vast datasets through complex neural networks. Engineers often cannot fully explain why a model produces a particular output. This gap is studied under a field called mechanistic interpretability.
India engages this emerging technology space mainly through international cooperation. India is a founding member of the Global Partnership on Artificial Intelligence (GPAI), launched in 2020 to guide responsible AI development. India was also a signatory to the Bletchley Declaration, the first multilateral statement on frontier AI risks, agreed in 2023 at the inaugural global AI Safety Summit in the United Kingdom. No single binding international treaty yet governs frontier AI systems. In this vacuum, private AI companies have built their own internal safety frameworks. One such framework trains models against a written set of ethical principles rather than relying only on human feedback, an approach known as Constitutional AI. Consulting religious and philosophical traditions on machine ethics is a recent extension of this same self-regulatory effort.
The alignment problem has two layers. The technical layer asks how to make a model reliably follow instructions and avoid harmful outputs. The deeper layer asks whose values a model should be aligned to, since different cultures and moral traditions disagree on core questions of right and wrong. Frontier AI companies have responded by trying to encode broad ethical principles, drawn from constitutions, human rights charters, and now religious traditions, directly into model training.
A parallel concern is dual-use risk. The same capabilities that make an AI model useful, such as chemistry, biology or coding knowledge, can also be misused to design weapons or conduct cyberattacks. …
Consider the following statements:
1.The 'AI alignment problem' refers to ensuring an AI system's goals and behaviour are consistent with human values and intentions.
2.'Mechanistic interpretability' research aims to understand, at a detailed level, why an AI model produces a given output.
Which of the statements given above is/are correct?
Match List I (Term) with List II (Description) and select the correct answer using the codes given below:
List I:
A.Constitutional AI
B.Bletchley Declaration
C.Mechanistic interpretability
D.Moral status
List II:
1.Training an AI model using a written set of principles rather than only human feedback
2.First multilateral declaration on frontier AI safety risks
3.Research field studying the internal workings of neural networks
4.Inherent entitlement to ethical consideration independent of usefulness
Consider the following statements regarding the Bletchley Declaration:
1.It was signed in 2023 at the first global AI Safety Summit held in the United Kingdom.
2.India was among the signatory countries.
3.It created a binding international treaty enforceable by a United Nations body.
How many of the statements given above are correct?
In philosophical usage, the term 'moral status' of an entity most precisely refers to:
Want the full note, Prelims Facts & MCQ answers?
Create a free account — takes less than a minute.
Create free account