As Bertrand Russell said, "Men are born ignorant, not stupid. They're made stupid by education". I wonder if they would stay stupid if the education continued as it does often through a well-lived life. But I think this way probably because I'm inherently hopeful about human sense-making of technology, tools and how we apply these in our specific landscapes.

If education today seems to have become about using ChatGPT to write your homework, we're manifesting this world view and breathing life into this specific idea of what education is – leading to a corruption of our minds, our work and our lives. But what if that was not the objective?

I once told Vitorino Ramos, one of the optimization researchers I know from X/Twitter, that life is like a particle swarm optimization with no "global best". In the Shanti Parva of the Mahabharata, there's advice for kings to learn many times, from many seers – "Bahuda shrotavyaha, bahubhir shrotavyaha".

The Abundance Mindset

What if humans could expand the field of possibilities to be huge, vast, and more all-encompassing, and the abundance of this befits the meager AI tools we have developed today, whether LLMs or diffusion models or JEPA or anything else a few years down the line? There are mathematical truths about them (such as the universal approximation theorem of deep neural networks), but these represent vertical slices of realities we contend with along the journey. These could be grounding mechanisms or constraints on the broader journey but perhaps never all-encompassing.

It is not hard to remember how engineers, thinkers and problem solvers a few decades ago were, X lingo, "jacked". I don't mean Jared Vennett jacked, but Kelly Johnson jacked – using slide rules to design the SR-71 or the U-2 spy plane, or used something with less computational power than a modern toaster to fly a spacecraft to the edge of Neptune and still receive telemetry from it in 2020.

And yet, at the risk of back-tracking from a steel man argument, one doesn't have to go handicapped through life, as is evident from how we can use Claude for one-shotting complex simulations and models.

The Purpose of AI-Infused Work

It follows that the purpose of education infused with AI in it, and work infused with AI in it is perhaps not in solving the mundane that has already been solved for the hundredth time. This capability exists to bring forth a struggle that is worthy of the capability, and my thesis from many thought experiments seems to converge on the need for a larger problem frontier.

I mean a ridiculously large problem frontier, the likes of which will make the current state of human endeavour and accomplish look mundane and trivial.

Thinking Bigger

It follows then, is that the big labs, the big tech firms and other AI doyens are probably not thinking big enough with AI – whether it the development of AI, or the use of AI. Consider that many labs have retreated to their lion's den of selling you ads via chatbots now rather than have them do useful things – this regression is deemed essential because of the discomfort that massive change means for incumbents.

And while it is easy to be challenged on the question of how big to think (surely, Rajesh, you don't think you know better than the big labs and these folks with fancy PhDs?), and it easier even to get the answer wrong (Rajesh says a specific thing in response, which falls flat). But consider that I am only a mirror for the many voices of people using AI in its current state – the signal is inescapable, and the inescapable conclusion is both for question asker and answerer, that the frontier of problems we use AI and human ingenuity for, needs to be widened and broadened.

I'm sure the reader can think of transformative tech such as electricity, powered flight, spaceflight and the rest. Some changes and some tools are so powerful that they change our world-view – and this behooves us to expand it.

Final Thoughts

And these are the words of an AI and technology optimist, regardless of the state of candor or irritation you may find me in at any point in time, due to any specific interaction with or use of AI tools. Frontier expansion is not a new idea, so I'm not being radical – only acknowledging the truth of the current state of things, technology, AI and the rest of it.