This song was included on Spotify’s Navidad en México playlist, and I guess its middle verse could sort of be interpreted as being about Christmas gifting, so I’m considering it in scope for this newsletter. What comes to mind for you when you listen through the dreamy, sparse lyrics?
Lyrics // My translation
Veo las luces cambiar en la ciudad // I see the lights change in the city
Hay robots y tengo que frenar // There are robots, I should pump the brakes
Creo que esto va a empezar // I believe this is about to startTodos buscan algo para dar // Everyone’s looking for something to give
Que en febrero, ya no vas a usar // That, by February, is already past its use
Tragedia internacional // International tragedyY yo ya sé no va a volver // And I already know it won’t return
Ese lugar que un día fue // That place that was before
Sin importar voy a intentar // No matter what, I’m going to tryEs otro año más // It’s another year
It makes me think about artificial intelligence. I don’t know how much of that was that this year was a crossover in my own experience of AI.
At the start of the year, I had started to use LLMs as part of my work, but didn’t find them all that helpful. Somewhat towards the middle of the year, I found that the AIs were now better at coding than me; not just providing useful tips and boilerplate code, but often proposing better designs than I would have come up with. I saw therapists less often because AIs have become very good at helping me to conversationally parse through my feelings. While writing these newsletters, I have used LLMs for things like pulling interesting quotes from large PDFs and answering specific questions (“does a two-track recording mean two physical tapes?”).
The LLMs have not written any of the sentences in these newsletters, because they are still worse at writing than I am. My translation above was machine-assisted, but some of the phrasing suggested by Google Translate / Claude was ugly and overly direct. Still, I think of Roger Ebert’s advice to “Try to get to the point where you write better than anyone who writes faster than you, and faster than anyone who writes better than you.” I am better, but they are so much faster.
Since June 2023, LLMs have gone from “no better than chance” to “as good as PhD students” at answering a set of multiple-choice questions written by domain experts and designed to be Google-proof. In December, I saw a Stanford professor give a talk where he noted that he now requests that his PhD students use LLMs to assist them in summarizing research (they turn in better summaries) and that general AI models now do better than doctors at assessing case histories. How much do these benchmarks (few-minute knowledge recall, few-hour programming and medical tasks) matter for automating labour? I’m unsure, but it makes me wonder whether I will be past my use soon; will knowledge work end up, to borrow Maxwell Tabarrok’s examples, more like “computers” after transistors were invented or ice harvesters after refrigerators were invented?
Some of my friends believe that we’re going to lose control of AI, essentially a new intelligent species that we’re partway through inventing, and end up in a world that is much worse for humans, perhaps wholly uninhabitable. Even if humanity muddles through, it feels like the near future is going to be made weird by intelligence on demand.
Right now, I mostly relate to LLMs as “my very unusually knowledgeable pal who is not above making things up in order to feel helpful, but who happily answers any of my questions” and “a way to journal with constant prompting”. But I don’t think those ways of relating are going to hold for very long. Will I feel watched over by machines of loving grace? Will I feel annoyed by the creeping shitty non-humanity infused into my daily life? Will violence get amplified by AI to the point where it ruins the earth for humanity, or even ends us? What should I be trying to do, given all my uncertainty?
I like that the second-to-last line of the song ― sin importar voy a intentar ― is completely vague about what she will be trying. Just, going to try, no matter what.
We’ll see where we’re at in another year,
Tessa