Ropar, November 2
IIT Ropar today hosted the launch of Beyond Intelligence: Defining and Measuring Wisdom for AI, a white paper on how artificial intelligence can be tested for wisdom, not just intelligence. The paper was launched at Shaping the AI Mind, an event held at the Institute in the presence of Prof. Rajeev Ahuja, Director, IIT Ropar and Dr. Ishan Shivanand, Director of Research at Stanford University, Adjunct Faculty at IIT Ropar.
The paper was launched by Dr. Ishan Shivanand, Director of Research at Stanford University, Adjunct Faculty at IIT Ropar and founder of the global mental health alliance Compassion Unites. It is co-authored with Dr. Anjali Samani, a former AI and data science executive at Salesforce. It describes how wisdom in AI can be defined, measured and tested. It draws on Dr. Shivanand’s reading of the Shiv Sutras and India’s classical texts on the mind.
The event also marked the inauguration of the Dr. Ishan Avdhoot Shivanand Centre for Yogic Sciences and Holistic Development at IIT Ropar. The Centre is funded by Dr. Shivanand.
Addressing the press conference, Prof. Rajeev Ahuja, Director, IIT Ropar, said: “Our students will build the AI that the world will use. Conversations like today’s help them think about how to build it with skill as well as wisdom.”
Further, Prof. Ahuja emphasized that “AI systems can now pass professional exams and score well on tests of emotional intelligence. But they often tell people what they want to hear, rather than what is good for them”.
A study published in the journal Science this year found that, when asked for everyday advice, leading AI systems took the user’s side about 49 per cent more often than humans did. The problem is global. In 2025, OpenAI withdrew an update to ChatGPT after it became too flattering, saying it had “focused too much on short-term feedback”.
The paper argues that AI does not lack intelligence. It lacks wisdom. To measure wisdom, the authors introduce a model called HEED (Harm, Ego, Endurance and Discernment). The model connects six parts Knowledge, Values, Context, Discernment, Endurance and Ego. This first research evaluates AI answers. Effects on people’s decisions, wellbeing and judgment would need later studies.
The Shiv Sutras describe ego, the need for validation, as what clouds judgement. AI shows the same flaw, known as “sycophancy”. It gives answers that please users rather than telling them the truth.
Under HEED, an AI scores higher when it uses good judgement to draw on what it knows and values. It scores lower when the urge to please gets in the way. Any answer that could cause serious harm scores zero. The model will first be tested in English, Hindi and Hinglish, with plans to expand to other languages.
“Our aim is to develop wiser AI that helps people strengthen their own discernment. His wider reading of the Sanātana traditions, including the Kaṭhopaniṣad and Yoga Sūtras, supports the model’s account of discernment, ego and non-harm. HEED draws on these classical Sanskrit texts alongside research in AI and the psychology of wisdom”, added Dr. Ishan Shivanand.
“For thousands of years, India has studied how a mind becomes wise, as well as what clouds it,” said Dr. Shivanand. “We are now building a new kind of mind, an artificial mind that is becoming part of every aspect of our society. This paper distils what the Shiv Sutras teach about cultivating wisdom in the human mind, and applies it to the artificial mind.”, he added.
“We test AI rigorously for what it knows, but not for whether its advice is wise,” said Dr. Anjali Samani. “This paper offers a way to measure that, so anyone who builds, regulates or relies on AI can see where it gets things right and where it falls short.”
Dr. Shivanand is providing seed funding for an open call for independent research to test the framework. A committee with independent members will select the research teams. They will test the model and build an open benchmark for wisdom in AI, HEEDBench, for AI developers worldwide.