That Made My Week 1 – Artificial Intelligence and Analogy

What is “That Made My Week?”

A weekly newsletter.

A place for me to share the best thing I’ve come across that week.

Something that means a lot to me. Maybe an idea, an article, or something inspiring someone is doing.

Something that conveys a sense that “Wow, I didn’t realize that was possible!”

Why?

I like to interact with people and brighten their lives. Writing about awesome things gives me the opportunity to experience them more deeply.

Me Write Code

It’s been awhile since my last blog post, though I have been writing a ton. I have also been diving into a different sort of writing entirely.

Following a turn of inspiration, I find myself learning software development. I took this on full time since last November. I attended bootcamp at a place called Codefellows from December until the end of May. I am now diving into the next phase — job-getting, leetcode, working on projects.

It’s had a result on my writing that may or may not be tangible to the reader. It feels like I have been doing resistance training. My thinking feels more transparent, sharper. More direct.

I'm curious to see how this pans out in the months and years ahead.
On the backburner: Retreats on the topic of making pivots in life. Creative retreats for those working in domains beyond writing.

Fun fact: “Computer Science” used to be called “Communication Science.”

… And now for the first edition of TMMW.

AI, Analogies, “Understanding”

https://www.quantamagazine.org/melanie-mitchell-trains-ai-to-think-with-analogies-20210714/

What defines artificial intelligence? What does it mean for the future of humanity?

This inspiring article is about Melanie Mitchell, a computer scientist. It was written by John Pavlus in Quanta Magazine. It explores why analogy-making is crucial for developing a truly intelligent AI.

Analogy allows our intelligence to transfer from one context to another. To make mental leaps. To think figuratively. To be reminded of things during the flow of a conversation. Analogy allows us to get to the gist of something. To summarize, to tell a spontaneous narrative, to put an experience in our own words.

These are beautiful human abilities. They express something of our individuality and also our sensitivity to context. The book you write is different from the book I write. You wouldn’t describe a memorable sunset in exactly the same way twice. We cater our communication to the circumstance and our listener. Analogy allows us to pivot based on what’s alive in the moment. More broadly speaking, analogy is something that stems from who we are rather than being defined by the task at hand.

Prof. Mitchell’s stance (following Turing and Hofstadter) is that for thinking to develop, it needs to happen in stages. Intelligence must build upon itself as it does for us as humans, not in the brute force black box style that happens with current paradigms of machine learning.

For artificial intelligence to be intelligent, we need to understand intelligence.

How can we account for the nonlinear leaps that bring about the experience of understanding? What of instinct and visceral intelligence, of gut level hunches, and of metacognition, or of knowing what we do and do no know? The ability to think in vague flights that can emerge into sparks of new ideas? What of emotional intelligence, which is empathic and subtle? What of the moral compass and a person’s quest for truth?

All this is important to me not only because I believe it will allow us to develop better AI. I want the AI we develop to be human-centric, to make our lives more enriched, meaningful, good, beautiful, true and spiritually vital. But this ambition this takes considerable effort and sensitivity. Unless we specifically set humanist AI as our aspiration, by default we move more towards the alternative, an AI that brings us towards materialistic reductionism and social injustice. It’s important for us to learn about the biases we have and also to put a great deal of care into the kinds of major strides we make with technology.

Another cool takeaway from this that I can’t help but mention has to do with how we categorize “computer science” as dealing with computers. It doesn’t have to get so directly compartmentalized. I would prefer the arrow of meaning point towards a grander mystery and also back at ourselves. Let’s celebrate computer science as it is, and also regard it as wizardry and, more broadly, “communication science.”

The article is a great read, and I hope it inspires something for you. Check it out!