The development of Artificial Intelligence (AI) that can learn and develop skills independently of its programmers has been a long-standing concern among scientists, ethicists and science-fiction writers. A recent interview with Google executives has added to these worries.

During an interview on CBS's “60 Minutes” on April 16, James Manyika, Google's senior vice president for technology and society, revealed that one of the company's AI systems had taught itself Bangla, a language it had not been programmed to know.

He stated that "with very few amounts of prompting in Bengali, it can now translate all of Bengali."

This raises concerns about the potential for AI to develop skills and abilities that its programmers did not intend.

Google CEO Sundar Pichai acknowledged that there were still aspects of how AI systems learn and behave that experts did not fully understand. He described this as a "black box" and admitted that the company needed more research to fully comprehend how it works.

This lack of understanding raises questions about the safety and reliability of AI systems, particularly as they become more powerful and capable.

Despite concerns about the risks of AI, Pichai believes that developing more robust safety layers before deploying more capable models can help mitigate potential dangers.

He has called for wide-ranging global regulation of AI, a view shared by other tech leaders such as Elon Musk.

Chinese lawmakers have already implemented new rules, while in Europe and the US, the regulatory process is still in its early stages.

The development of AI has not been without its flaws, with examples of fake news, deepfakes and weaponization.

Pichai acknowledged that all AI models suffered from "hallucination problems" – where the system generates false information or outputs.

These issues highlight the need for increased regulation and safety measures to ensure that AI is developed and used ethically and responsibly.