[FRIAM] AI art

[Barry MacKichan]

I was struck recently by an article in *Harvard Magazine* by Harry 
Lewis, a computer science professor and former dean of Harvard College. 
In it, he says:

“The first error is suggesting that computers can be digitally trained 
to be superior versions of human intellects. And the second is inferring 
that human judgment will not be needed once computers get smart enough.

“Learning first. How do we become the people we are? Not by training a 
blank slate with discrete experience data. Our brains start off prewired 
to a significant if poorly understood degree. And then we learn from the 
full range of human experience in all its serendipitous contingency. We 
learn from the feel of an embrace and taste of ice cream, from battle 
wounds and wedding ceremonies and athletic defeats, from bee stings and 
dog licks, from watching sunsets and riding roller coasters and reading 
Keats aloud and listening to Mozart alone. Trying to train a computer 
about the meaning of love or grief is like trying to tell a stranger 
about rock and roll.

“And not just from all these life experiences of our own but from the 
experiences of all our cultural ancestors. Our teachers—all those from 
whom we have learned, and those departed souls who taught our 
teachers—have shaped all those experiences into a structure of life, a 
system of values and ideals, a way in which we see and interpret the 
world.”

Some friends from my college days have shared a quiz of mathematical 
puzzles. I gave the quiz to Github Copilot, since that LLM seems well 
versed in mathematics. The LLM aced it, and in addition, identified the 
theorems that allowed it (him? her?) to solve the problems. This is 
somewhat surprising given that LLMs have a hard time counting to four 
(One reviewer asked ChatGPT 4o to come up with a four-syllable word 
beginning with “w”. The answer it gave was “wonderful”.)

Concerning mathematics, my take on it currently that it can and does 
master learning existing mathematics, and it conceivably can draw 
analogies and make connections between the various fields of 
mathematics, but I do not see that an LLM can create genuinely new 
mathematics; I can’t see how an LLM trained in 1858 could possibly 
state the Riemann Hypothesis, let alone prove it.

My current opinion is that if you think of mathematics as a bunch of 
points in a high dimensional space, AI can fill in the spaces between 
two of those points; i.e., it can find new results, but it cannot head 
out beyond the existing points. In mathematical lingo, it can’t go 
outside the convex hull of existing mathematics.

— Barry


PS
For renegade English majors, a convex set has the property that if any 
two points are in the set, then all the points on the line connecting 
them is also in the set. A ball is convex; a doughnut is not. The convex 
hull of a set is the smallest convex set containing the original set. 
The convex hull of the doughnut is what you get when you completely fill 
in the hole.

On 22 Jun 2024, at 23:32, Pieter Steenekamp wrote:

> AI will never fully replace humans in the realm of pure art. While AI 
> has
> made impressive strides in generating art, humans inherently gravitate
> towards creations by fellow humans. This preference stems from our
> deep-seated connection to human experiences and emotions.
>
> Consider the analogy of chess: AI can easily outplay the human world 
> chess
> champion, yet we remain uninterested in AI-exclusive tournaments. The
> reason is simple—our fascination lies with human competitors and 
> their
> stories, not with machines. This extends beyond chess to all forms of 
> art.
> Whether it’s music, literature, or visual arts, the knowledge that a 
> human
> mind and soul crafted the piece adds a unique layer of significance.
>
> It's not that humans are disinterested in non-human phenomena such as 
> AI,
> the stars, or mathematics. There is a wide spectrum of interests among
> individuals, with some drawn more to human-centered pursuits and 
> others to
> abstract or scientific endeavours. However, as a collective, 
> human-related
> creations consistently hold a special place in our hearts.
>
> When a human plays chess, the essence of the game is enriched by 
> knowing
> the opponent is also human. Similarly, when we listen to music, read a
> novel, or admire a painting, the awareness that it was created by 
> another
> human being adds depth to our appreciation. This connection to the 
> human
> aspect of art is, in my opinion, irreplaceable by AI.
>
> I can't prove this definitively; it is simply my perspective.
> -. --- - / ...- .- .-.. .. -.. / -- --- .-. ... . / -.-. --- -.. .
> FRIAM Applied Complexity Group listserv
> Fridays 9a-12p Friday St. Johns Cafe   /   Thursdays 9a-12p Zoom 
> https://bit.ly/virtualfriam
> to (un)subscribe http://redfish.com/mailman/listinfo/friam_redfish.com
> FRIAM-COMIC http://friam-comic.blogspot.com/
> archives:  5/2017 thru present 
> https://redfish.com/pipermail/friam_redfish.com/
>   1/2003 thru 6/2021  http://friam.383.s1.nabble.com/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://redfish.com/pipermail/friam_redfish.com/attachments/20240624/2dc30422/attachment.html>