Today I was walking across a hospital corridor while my father was undergoing dialysis. Along the walls, there are a couple of images laid out very neatly in wooden frames.
If you look at them clearly, these are not paintings. They are images printed on canvas, and that is quite fascinating, because they are very highly detailed. They have probably been treated for the hospital too, and wiped down with alcohol more times than I can count.
It really amazes me that we have technology that can print high-quality images on canvas. At the same time, we have black and white printers that don’t work about one third of the time, and we get frustrated with them.
This reminds me how much innovation it would have taken us to go from writing by hand to the printing press, then to inkjet and other printers, and then to this one in particular, where we can print really good images on canvas.
If I do a mental frame of it, I go all the way back to where we arrived at writing. The earliest tablets come from Uruk, in southern Mesopotamia, the same city where Gilgamesh would later be king. People pressed signs into damp clay to record things like rations and goods. From those clay tablets to a canvas print in a hospital corridor, it’s like a complete history of civilization.
Even if you just take the advancements in printing alone, those are quite baffling. Really good stuff keeps coming out. But imagine how much time someone would have spent on trial and error to arrive at: hey, this thing actually works with 90% accuracy, and it fits what we need. Every one of those prints on the wall is sitting on top of thousands of attempts that didn’t work.
Now imagine you have something that can actually help you speed up that trial and error. That’s what I’m hinting at with AI. It can literally come up with combinations and recipes, take in all of the data and other fascinating things, and speed the whole process up, so that we arrive at amazing things like these prints, or maybe even better ones.
The research alone is moving at an unprecedented rate. According to Stanford’s 2024 AI Index, the number of AI publications worldwide nearly tripled between 2010 and 2022, from about 88,000 to more than 240,000 a year. And that was before most of the current tools even existed.
Not all of this will be correct. I’m not saying all of it is useful. But people are trying things out, and knowledge is going from our heads onto paper, which could be useful. It gives people a way to speed up trial and error, and we can produce good results. AI can help us out in a lot of ways. It just depends on how we want to use it.
This isn’t a big essay. It’s just a thought that was in my head as I walked through the corridors of the hospital, and I wanted to put it out there.
On the way back, I passed the same frames again. Clay tablets in Uruk on one end, a canvas print on a hospital wall on the other, and somewhere in between, thousands of years of people trying, failing, and trying again. That’s a nice thought to walk with.
Sources
- Cuneiform, an introduction, Smarthistory
- AI Index Report 2024, Chapter 1: Research and Development, Stanford HAI
