When Machines Learn Faster Than Us

The danger lies not in machines getting more intelligent with time... but in humans finding it convenient to depend on that intelligence.

- D. Banerjee and Sourav K

Not so long ago, learning was something only humans could do. Children learnt through textbooks, and professionals took decades to master their skills — the road to expertise was long and filled with mistakes and mentorships. Today, however, a machine can completely master in mere minutes what a human would take years to learn. A generative AI model can write code that it was never explicitly taught, diagnose diseases from scans better than trained specialists, and create marketing strategies without attending business school.

Artificial intelligence is not arriving with a big bang or in the form of dystopian robot armies. It is coming in a subtle way—through recommendation engines, chatbots, automated hiring systems, predictive analytics, and decision-making algorithms that are deeply embedded in everyday systems. And while AI is not ‘killing’ humans in a literal sense, it may be taking away what makes human contribution economically and cognitively relevant. Are we witnessing progress—or the slow outsourcing of human thinking?

From tools to thinkers: How AI crossed the line

For decades, technology acted as an amplifier of human capability. Calculators enhanced arithmetic. Computers accelerated computation. Software automated repetitive tasks. Humans remained firmly ‘in the loop’.

Today, AI systems have not only come close to the level of human intelligence but have also surpassed it in many ways. Besides carrying out commands, they now analyse, foresee, and improve their performance. Large language models can reason in different fields. Vision machines can sense patterns that are not visible even to the human eye. Reinforcement learners are becoming better with the help of the loops of feedback in a very short time—far quicker than any human could ever be.

In 2023, researchers noticed that AI models were able to do tasks better than humans in:

• Diagnostics of medical images