Hey folks,

If you’ve watched any of my livestreams when someone asks why I don’t get ChatGPT or something to do a task for me, you probably saw a pained expression come across my face. Part of me dies every time someone tells me that they used some LLM chatbot to solve a problem. I have many reasons for despising the fascination with AI-based tools. I even wrote a commentary that I submitted to mBio in the fall of 2024. Yes, I wrote it. By hand. Then I typed it. No really, I typed it on a typewriter. My thought was that there’d be no mistake that the essay was written without the use of AI. When I went to submit it, I really wanted to mail in 5 copies to the journal, but didn’t know where to send them. Alas, I needed to use my phone to take pictures of the typed pages and convert them to a PDF. Sadly, it didn’t get accepted. I periodically think about trying again. If anything, my arguments have only gotten more numerous and entrenched. Whatever. I feel like I’d just be spitting into the wind. Here’s my basic argument: AI is dehumanizing. You deserve better because you are better. If you want me to unpack that let me know.

I’ve become fascinated by attempts to decipher whether content was AI generated. Some cases are pretty obvious. It’s become somewhat comedic to see reactions to AI-generated content whether it’s books, music, or images. In these cases, I think much of the backlash is the lack of disclosure that AI was used. Audiences feel duped and that there has been a breech of trust by the content creator. I don’t see the problem getting any better. In fact, I suspect it will get worse as the models improve and the Overton window expands.

So what to do about it? Well, I can think of a few strategies.

The fist is to look for AI “tells”. Things like Taylor Swift lip-synching or having super human perfect pitch. Or the 6-fingered blues guitarist from the 50’s no one knew about until last year. Or a rat with oversized genitals. Meh. Sure, these are fun detective stories, but these tells can only get us so far. Also, AI detection tools generate obvious false positives. I suspect the tells will slowly go away. At the end of the day the “artists” have to live with themselves. Increasingly, their consciences aren’t troubled.

The second strategy is to engage in and consume media where the likelihood of using AI is minimized. For example, my handwriting and manually typing that commentary. Consider also my livestream videos. You get to see all my mistakes, troubleshooting, attempts to respond to comments and questions, pivots in the direction I take a visualization. Basically, the strategy is to increase the physicality and personal interaction inherent of the experience.

Related to the first and second strategies is a third. Get really freaking good at your craft. Do it better than everyone else, better than AI. I suspect AI would not have helped Shakespeare. In fact, it probably would have hindered him. So much of what I critique are data “visualization memes” like stacked bar plots and dynamite plots. Visualizations that people generate without thinking because everyone else is. That’s the kind of “slop” that AI can generate as well as humans. I haven’t struggled to find examples of human-generated data visualization slop to worry about whether it’s AI generated slop

How do we hone our craft? Practice. Critique. Repeat. I have become passionate about helping others improve their craft when it comes to data visualization and reproducible research. If you’re interested in working with me on an individual or group level let me know. I am offering consultation sessions focusing on improving your data visualizations. If you are interested in learning more about what I can provide you, please sign up for a free 30-minute exploratory meeting.

If you would like some more practice with critiquing data visualizations, check out this figure that I just found in the journal Nature Cell Biology titled, “Characterizing the metabolomes of microglia, astrocytes and neurons in ageing and Alzheimer’s brains”. My challenge to you is to make a list of things you like in this set of panels vs things that are obviously problems - dare I say “slop”. Once you’ve got a list, compare it with what your friends came up with.

​

​