I Wish We Would Stop Calling This AI
This is not AI. We'll know when it's actual Artificial Intelligence.
A few weeks ago, I learned that Meta’s AI was trained on 3 of my books. If you’re an author or an artist, Meta’s AI was probably trained on your work as well. It’s not an exaggeration to say Meta stole our work and intends to make money off of it. And this isn’t the first time this has happened. My writing has been stolen and used to train other AI models as well.
But it didn’t have to be this way, and an argument could be made that it’s not too late to change things.
Let’s talk about generative AI tools.
This Isn’t AI
First of all. This isn’t AI. Actual Artificial Intelligence will change everything in ways we can’t predict or imagine (more on that later).
The generative “AI” being shoved down our throats at the moment is just another new tech tool, one that people may or may not find helpful.
To be clear, I understand a lot of people really enjoy using it. I understand some people find it incredibly helpful. I understand it can feel transformative for certain kinds of work. I understand that sometimes the results of an AI query can feel magical. I too have witnessed that magical feeling from an AI creation.
Though I should admit: I’ve never made an AI query myself. I tried once. It was probably a year ago. I have several family members who use AI regularly and encouraged me to try it out when I was stuck on a task. So I tried. But my query happened to coincide with ChatGPT being unavailable — a rare occurrence I’m told. After spending some time trying to figure out what I’d done wrong, I realized this tool wasn’t helping me, and was literally wasting my time, and I went on about my work.
And still, my lack of personal AI use has not enabled me to steer clear of it. It’s everywhere. It’s on platforms where I don’t want it or need it. And it seems impossible to opt out. It’s obviously not going away.
But it’s not intelligence. These AI platforms are not thinking their own thoughts. They are just compiling and regurgitating pieces of their database. And they get things wrong as often as they get things right. In art, that means they do things like add too many fingers. In writing, it means that spew nonsense. My favorite recent example is that someone discovered you can type any nonsense phrase into Google’s AI, and add the word “meaning” at the end, and it will regurgitate an explanation of what the non-existent saying means. Here are two examples:
Search: under the tortilla fortress meaning
AI response: "Under the tortilla fortress" is a phrase that can be interpreted in a few ways, but it generally implies a sense of being sheltered, protected, or hidden within the confines of a tortilla. It can be a metaphor for:
Privacy or seclusion: The tortilla acts as a barrier, creating a sense of being out of reach or shielded from outside influences.
Comfort or warmth: The tortilla, being warm and soft, can represent a sense of security and comfort.
Search: one for the toenail, two for the shoe meaning
AI response: The saying "one for the toenail, two for the shoe" is a shortened version of the proverb "for want of a nail the kingdom was lost," which emphasizes that seemingly insignificant details can have major consequences. It highlights the importance of attention to detail and the potential for small errors to lead to larger problems.
[You can see more examples of these nonsense phrases in this post on Threads.]
The written responses that AI tools generate sound authoritative and factual, so it’s easy to get duped by them. But the answers they generate can’t be trusted. In my current college class (I’m enrolled in the philosophy degree at Newlane), the professors advise that it’s fine to use AI tools to brainstorm ideas for a paper, but that we need to be careful because AI tools will also cite books and articles that don’t actually exist. Yikes!
Ben Blair uses AI tools to help evaluate documents and says the AI routinely misses obvious facts. And when he asks the AI: Why didn’t you include this listed item? The AI will simply respond: Yes, that item should have been included. But it gives no reason for not including it in the first place. Ben also says that every time the AI tool shows the total number of items, it has done the math incorrectly and has the wrong number for the total.
Because again, the AI tool is not thinking. It’s not actual intelligence. It’s just aggregating and editing and presenting pieces of its vast database that might relate to your query.
It’s Not Intelligence, It’s Just Another Tool
I find it helpful to compare this new tech tool to other tech tools that have changed things on a grand scale. If you’re at least as old as I am, I think a good comparison is email. It’s hard to describe how magical email felt when it first became a regular part of our lives. I was a sophomore in college in 1993 when I sent my first email message — written from my campus computer lab — to a friend studying in Jerusalem. I don’t pretend I understood at that moment how transformative email tech would be to people’s lives and people’s work. But I definitely understood that it felt special and extraordinary. Just like the way these AI tools feel today.