Before the recent AI buzz, we had ML

Before the recent AI buzz, we had the ML (Machine Learning) buzz. I believe thinking in terms of the latter is a better way to assess the long term usefulness of the technology.

My friend Thanh, who is working on building solutions that involve AI powered code generation shared with me how he thought of the system he is building:

i think of the AI as an intern … eventually the system should know enough to work independently

I thought this was very interesting. In this article i muse on that thought and share some related ideas.

At the heart of the discussion is the issue of learning (these days it is easy to forget that before “AI” became hot, “ML” was all the rage - machine learning !¹).

Going back to Thanh’s analogy, an intern has the inherent potential to learn. This means, one day we can fully expect the intern to acquire the ability to get things done independently. Is it reasonable to expect the same of AI agents ?

How do we even approach this question ? How do we evaluate if AI systems can effectively learn ? I mean, we have no good explanation for the learning ability of humans so that we can look for ways to replicate it.²

As far as we know, humans have the ability to learn at birth. What education or training does is merely to exercise that ability and hone it.

So, how do we best describe the role of an engineer building an AI powered system ? I think, the role of a teacher or a trainer is what best fits.

After the engineer teaches the system for a while, the system is then released into the “real” world, just like students graduate from college and find jobs. In both the cases - AI systems and students - we can only hope that the inherent learning ability present in them to carry the day.

The only difference is that, while students of every generation have proven that ability, we are yet waiting for the first AI success story !