Thoughts: LLMs - A Weekend Drive and a Steak Dinner
I was having a chat with a friend recently which kept circling around LLM (or "AI") usage within our respective jobs, how people around us use them, and our own feelings about engaging with LLMs.
We came to a conclusion that actively using LLMs is a lot like eating steak: something many people will openly acknowledge as being "bad" while still willfully (often even eagerly) partaking.
The eating of a steak is arguably morally evil on multiple levels: the killing of an animal for food, the funding of an increasingly cruel animal farming industry, and the side-effects of cow farming on the global environment. These are all things most semi-empathetic adults can agree are bad, but these badnesses are all far away compared to the immediacy and enjoyment of "Yum, steak!"
A similar comparison is the driving of personal gas-powered vehicles, especially when other modes of transport are viable: an activity that very literally pays money to oligarchs, pollutes the planet on multiple fronts, and encourages anti-pedestrian city design -- but goddamn it's raining a little and the grocery store is a 30-minute walk; complete environmental meltdown isn't going to happen today.
Using LLMs, eating steak, and driving cars are all enjoyable, convenient, and harmless to most peoples' immediate surroundings. I don't know that the people who actively partake in or enjoy these things (myself included) are inherently evil, even if the companies and philosophies funded by these activities most definitely are. Maybe a bit evil?
I think the conveniences or pleasure afforded by these activities are simply too great for many when weighed against the nebulous but obvious far-off penalties. Out of sight, out of mind. Yet I do think even those who "blindly" enjoy these things are cognizant of the inherent negativity of their deeds -- one only need look as far as their defensiveness when confronted with those who have eschewed these comforts. The angry annoyance at veganism, the teasing of a colleague who rides their bike to work even in the winter, the vehement dismissal of anti-LLM voices as "falling behind". These are not the reactions of clean consciences, even if the guilt is buried deep or never acknowledged.
And then there are those of us with a particular combination of privilege, education, and laziness to know exactly how bad the things we do are, and yet do them anyways. All of the guilt, no effort at reform.
Here I bear my own guilt as an eater of meat, a driver of cars, and an occasional orchestrator of LLM solutions. My family, lifestyle, and professional role ensure these are all things I likely will never escape. Inertia's a bitch and there's mouths to feed. I take the small joys and conveniences offered to me in life for a few bucks a pound, a dollar a litre, and a cent per 10k tokens, all paid to the monsters devouring our planet.
Sitting here, having written all this, I honestly couldn't tell you what what it would take for me to change these parts of my life. I'll generally keep AI out of non-work projects, I'll opt for pescetarian or vegetarian dishes when the options are likely equally enjoyable, I'll ride my bike to the store if it's really nice out instead of driving, but these are all fairly low effort changes that take next to no real commitment on my part.
It's weird to admit that I'm part of a problem and likely won't put in the effort to change, but there it is. I'm practically a cartoon villain, here. Perhaps this self reflection can be a small step towards trying a bit harder.
So while I swallow 1/8th teaspoon of guilt as seasoning on my steak and sheepishly have AI generate me an accessible typeahead search-select input to save 30 minutes at the office I drove to, I quietly applaud those with the willpower or moral fiber to actively and consciously decide to choose conscience over convenience, even for the little things.
If they're sometimes asses about it, we honestly kinda deserve it.