#193 - I Maketh Technology For Thee, Not For Me
On Artificial Intelligence and other tech you should stay away from.
Practice Drawing This 3D Model. The last one was false, so I can harp on that.
This is why AI is not going to replace artists: it uses a very fixed process where it collages things it found online, which is just not how creatives work. We use many processes, and the way AI works is just one of them.
There is this mantra, “process, not product,” which means you should focus on the “process”, and the “product” will sort itself out.
I didn’t want to be the one telling tech bros they got that part wrong. After all, they are focusing solely on the “product,” images, and texts. It looks legit, but you can often tell that humans did not make it, especially when it counts.
But they already know.
Sam Altman
Sam Altman, co-founder of OpenAI, which created the whole AI craze, does not use AI to work through ideas!
Get this: He uses notebooks, with pens I use, too, like Uniball Micro and Muji pens. They’re great pens. He uses notebooks small enough to carry around. I have a different one—it’s a bigger sketchbook, but I also carry it around to work through ideas.
See for yourself in this video here.
That is why I now feel safe writing this. I’m not alerting the tech bros that the creative process differs significantly from how AI works. They already know. Sam Altman has a creative process that’s very different from how AI works. He writes in notebooks.
Example: My New Captcas
To give you an example, this is what I designed recently.
I encountered a problem: hackers used my newsletter subscription form to spam newsletter join requests to arbitrary email addresses. They can do that for several reasons. One, they may try to get my newsletter blacklisted as they let the emails be sent to email addresses designed to catch fake emails. Or they could fill a finance department’s mailbox so it won’t notice hackers doing financial wrongdoings to that company.
I started getting lots of subscribe requests! Initially, it was great. But then I noticed things were off. I got a subscribe request around once every ten minutes, precisely. And it was always from the exact same form on the page, of the many forms I have in many places. And the people never completed the forms. When I checked, the email addresses seemed to have nothing to do with people wanting to learn to draw.
I took down the subscription forms and went to work on a captcha. I like to code everything myself. In this case, the advantages were that I could develop a new type of captcha that hackers had not trained their systems on yet and make the captcha “on-brand,” looking a bit like my line art.
Here are some snippets from my sketchbook where you can see me working through ideas on how to draw letters in lines, with a background consisting of lines, so humans can easily read it but computers can’t. My wife’s idea was to break lines open. This prevents hackers from finding the letters by just filling areas with a color.
These preparatory sketches are collages of existing shapes, but there is a feedback loop where I evaluate and develop the ideas further or decide to take a different path. I make quick sketches and quick studies, look at them, try to figure out what’s right or wrong with them, and try new sketches where I strengthen what works or try new approaches when they don’t work.
This is what the final captchas look like.
It was fun to design it in a sketchbook, code it, and see it come alive.
And it worked! The captcha stopped the hackers dead in their tracks.
AI could never, ever have come up with this. For one, it can not design captchas that can only be read by humans because, by definition, it can not read them itself, and thus, it can not check whether they would work. We humans can see instantly. And it could never have come up with an original new way to do captchas because it works by collaging what already exists.
AI might have a place when it comes to doing things we already know how to do. But real problem-solving and finding novel solutions are still done in your humble sketchbook.
Steve Jobs
There is something wrong with technology if the creators of that tech actively avoid using it themselves.
Sam Altman himself doesn’t use AI to ideate. Why should anyone else?
The iPad is another example of this.
Steve Jobs, co-founder of Apple, which created the addictive technology called the iPad, famously did not let his children on those iPads.
In other words, these people created tech, realized the downsides of using that tech, and didn’t use it themselves!
Practice Drawing This Is Different
Practice Drawing This has novel new tech, too: it has parametric 3D models which allow for randomization, which means you have a different reference to use for drawing each time. These are generated, but I hand-code the generators in contrast to AI generators.
And there are the Art Flashcards which are unique to the site.
What I made is hopefully as addictive as the iPad, and I hope you agree that is a good thing.
The difference between me and the Sam Altmans and Steve Jobs’s in the world is that I use the tech myself. I try to draw daily and use my Art Flashcards and 3D models.
That is how I know it is not evil tech. I hope that by using hand-coded 3D model generators, I can make drawing as addictive as possible. I think that’s a good thing. We experience resistance enough to start drawing already.
What It Means To Be Human
I don’t even let AI write my articles. I wouldn’t let robots do my skiing for me, either. Why delegate the fun parts of life to bots?
Computers can beat humans in chess, but we look at world championships where humans play each other. Ferraris can outrace an athlete, but we look at athletes racing each other. We don’t add a Ferrari to the mix. This is because we subconsciously compare ourselves to these people. How is it to be them? Could we compete against them? No one asks if they can outrun a Ferrari.
There have been no New York Times Bestsellers or Oscar-winning scripts written by AI yet. Part of the reason is that storytelling is about sharing life experiences, and computers don’t live lives (!).
The AI collage process leads to impressive-looking results that are generic, bland, never radical, and never surprising. Impressive, but bland.
This Week
This Saturday, another video premieres on YouTube:
It plays from 4PM CET, and shows twisted boxes in various orientations.
You may have noticed that I chose a different timeslot for these videos. I did this because it covers more awake time around the world. For one, it starts on Saturday and ends on Sunday, two days when people are generally off. Timezone-wise:
- In Los Angeles, it plays from 7 AM.
- New York, it plays from 10 AM.
- Amsterdam, from 4 PM, people will have eight hours to watch part of it before bed.
- In Hanoi, it plays from 9 PM, so viewers have a few hours in the evening or morning to watch it live.
- Beijing, 10 PM, so either a few hours in the evening.
- Sydney, Australia, it plays Sunday from 1 AM, so they are probably asleep...
This schedule may maximize the amount of time people around the world can watch this video premiere live.
Check out this video on one way to go about drawing twisted boxes:
