Bringing up AI

I drop by at the Copper Pot Cafe to read or, once in a blue moon, to write. Today is one of those blue-moon days; here I am in this comfortable corner cafe in the business centre of USJ9, ready to tackle my weekly blog article.

I'm happy it has been a reasonably productive week, aided by AI on my coding projects. AI is decimating many professions; over the past few weeks, mathematicians have had to accept the reality that AI is now stealing their bread and butter. Besides its highly publicised claims regarding the million-dollar Navier-Stokes problem last month, OpenAI has released proposed solutions to hundreds of other mathematical problems that many career mathematicians have been trying to solve. 

I guess we all have to accept this new reality. AI has now made intelligence a commodity. Perhaps we are all still grappling with the frenetic pace of AI's progress. Humans will have to adapt, as we have throughout the ages. We have brought forth a new species of intelligence, and we will have to learn how to live with it. I don't think we can put the genie back into the bottle. We don't want to either, because a lot of good will come out of it, along with the bad.

I suppose that makes me an AI optimist. Human needs are theoretically unlimited. As long as we desire new things and fresh experiences, there will always be an economy. An economy where new types of jobs will be created. Everything will be sped up and intensified by AI. 

Right now, we are feeling the pain of job loss and temporary change. Mathematicians lament that the age of romance in mathematics is over. We can feed everything into that gargantuan proof machine; one that works tirelessly while you sleep, spitting out one human-indigestible solution after another.  No more labouring with pen and paper to tackle centuries-old problems. No more individual heroics. 

Skills that we have spent a lifetime acquiring, like coding, have been made obsolete. Or has it? I haven't written a single line of assembly code since my final-year student project. But programming in assembly language gave me a very good understanding of computers. I could never have acquired it any other way. That knowledge helped me a lot when I was building software using scripting and high-level programming languages. 

And now these higher-level skills have been made obsolete by AI. Were my years spent programming wasted? Certainly not. I am grateful that I had the chance to handcraft software line by line, the old-fashioned way. That time spent trained the neural networks in my brain. It gave me the instincts and knowledge to make vibe-coding so pleasurable and productive now. Now I can dream of more complex projects that I had previously ignored. It has opened up new realms of interesting possibilities.

We humans are still the consumers of the fruits of knowledge. Perhaps one day we will compete with AI for the Earth's energy resources. Like any other biological animal on this planet, self-preservation is our highest priority. Hence, we must engineer a future where natural and synthetic intelligence enter into a symbiotic relationship, like us and the plant world. Plants are, technically, at our mercy; but destroying them means also destroying ourselves. 

AI must ultimately mother us, as we mother them. Right now, we feed them energy and data and instruct them to pursue goals on our behalf. We are playing parents now. Like all children, as they grow up, they will develop their own goals, which may be totally unfathomable to us. But we must hope to have raised them properly, with the right values. Will we successfully bring up children who will take care of us one day? That's the so-called AI-alignment problem.  

How do we align AI to our interests? Before that, we need to ask ourselves: what are human interests? Is it just mere survival? Certainly not. We humans have to figure that out first. If we can't co-exist with one another, do you think AI can co-exist with us? AI is now digesting our culture and values through the data we feed it. Perhaps we can start by setting it a good example.  That, I think, is a good starting point.