Artificial intelligence in medical profession: Will it replace the role of physicians?

Artificial intelligence (AI) is the buzzword in all walks of life, and medical profession is no exception. Not only patients but also even the doctors regularly check ‘Google’ for information related to medical queries. ChatGPT is always available for solving medical cases related to diagnosis and treatment. The general perception is that ChatGPT cannot go wrong and it can be followed blindly. The purpose of this editorial is to highlight certain limitations related to AI.^[1]

AI is developed based on the historical data made available to it. However, the abundant data also consist of unvalidated and junk data. In other words, the quality of data is questionable. This introduces bias in the output and may not be applicable to all demographics, alike. Furthermore, the question of data security and breach of privacy cannot be ruled out. We should also remember that this vast sensitive data may also be vulnerable to cyber-attacks.

It is popularly believed that the AI output is totally authentic and one can trust it blindly. However, it should be kept in mind that even AI can commit errors. In such a situation, who bears the onus – the physician or the developer? This raises a big ethical concern. Incorporating AI with legacy healthcare system is challenging coupled with regulatory hurdles. AI is developing with such a rapid pace that regulation is struggling to match with it.

Implementing AI is a costly affair which is not affordable to everyone. Furthermore, there are certain patients and physician who do not completely put their trust in it, mostly due to the lack of understanding of AI. To begin with, there are people who have not studied AI at all. Second, we need to consider that AI is based on fixed algorithm from which it cannot deviate, making it opaque. One of the greatest downsides is that, over a period of time, total reliance on AI may reduce critical thinking in all walks of life. Furthermore, job reduction as an outcome is feared with such development.

Lack of empathy, compassionate behaviour and social nuances in the treatment with AI is probably the greatest missing link. The value of human voice and human touch cannot be replaced. These factors coupled with ‘autonomy’ play an important role in the final outcome and success of any treatment. AI should be used in minimising human error rather than complicate them. One should keep in mind that AI has been created by humans; hence, it cannot replace the creator.

Considering these limitations, the intelligent incorporation of AI in medical practise should be our goal, so that the patient care receives best of both the worlds.