Jay Hoffmann

Books, movies, and code


#29: I think I’m using AI wrong

I think I’m using AI wrong.

Not wrong, obviously. There’s no right way to use AI. In fact, there may not be an ideal use case and if there is we certainly haven’t found it yet. And, as Molly White pointed out, maybe we shouldn’t be using it all.

What I mean is that I think I’m using it differently. When I compare notes against how other people are using it, I seem to be a bit off the mark. For instance, I see a lot of people that use it to help them think. They field questions against a chatbox, or use it to explain a chunk of code for them, or to connect disparate ideas together. Stuff like that.

It makes sense, I guess. I can’t wrap my head around it. When I first started using AI a bit in my writing, I used it to revise and rewrite a few things, which is where I find I guess most stuck. But then everything spit out in that sing songy, hustle culture, corporate-speak-with-exclamation-points-on-every-sentence way and I just gave up.

When I turn to it now, which is less and less, I use it to finish a thought. If I’m writing a bit of code, I’ll map out the structure and start to write a function name, then let Copilot fill in a first draft for me to look at. If I’m looking for a certain word or a turn of a phrase, maybe I’ll take that over to ChatGPT.

When I have something on the tip of my tongue, I turn to AI. And I haven’t really found another way to use it.


12 Years a Slave (2013)

May 13, 2024

When I first saw Steve McQueen’s Hunger it almost immediately became on of my favorite films of all time. A debut feature that reckons with such an immensely important and fraught period of history. That comes right up against it, cracks it open, and shows you a full view of the historical moment through the lens of a single person. It beautifully uses camera movement, intimate visuals, and didactic sound that draws you in and makes you pay attention to each tragic and gut-wrenching turn.

And somehow, 12 Years a Slave never made it to the top of my list for too many years. But here he is, McQueen doing exactly the same thing at another period of time that is illogical and brutal and almost beyond comprehension. It is an intimate, profound and uniquely horrifying portrait of a single man’s struggle to survive. This is not slavery glossed over with historical romanticism (there is only one man, really, who stands against it, in fleeting moments at the end of the film). This is the full moment of history, expressed through one man, cracked open for us to see set against the majestic landscapes that only serve to starkly contrast against the brutality of slavery. It’s a tough watch, no doubt about it. But it is incredible.

Stray Observations

  • Brilliant use of sound and silence, especially when the camera hangs there and really lets you feel it
  • The use of closeups to show the anger and hatred of man and the wide to show tragedy 
  • There’s something really disturbing about juxtaposing the beauty of the southern landscape against the horror of slavery

May 11, 2024

Good Grief (2023)

May 9, 2024

Dan Levy is a great writer. I really don’t know yet if he is a great creator of cinema. But his writing is so intimate and sharp and poignant it’s really incredible to watch. This film is a slow portrait of grief, one that tries to recognize and acknowledge how fraught the process can be. It is not even, and it is not possible alone. And that, on its own, is more than enough to make a movie about. And I’m glad he did.

Stray Observations

  • – “I’m so sorry I stood in the way of your snapdragons” – fathers eulogy
  • “We are standing in a house of loss.” Monet

May 7, 2024

A lot of focus lately on the failures of the latest swatch of hardware startups jamming AI into their tiny wearable nothings (why these need to be hardware appears to be a simple question of margins. Very hard to charge for API requests to OpenAI, but if you wrap it in a “wearable” you can charge $500 or more).

Anyway, a lot of them have an Apple vibe. But they don’t actually do anything. And, as Benjamin Sandofsky has now pointed out, they don’t have the Apple ethos either.

I blame the team’s nostalgia. They clearly want to recreate the Apple from 2007, but that’s impossible under venture capital constraints and without the momentum of Apple. Contrary to what Imran, Ken, and I’m sure many others at Humane believe, the iPhone didn’t begin with their work in the 2000’s on Project Purple. It began in 1976 with the Apple computer, and the decades of goodwill it built up in consumers. The project was spearheaded by a guy ready to waste billions in iPod revenue if it helped achieve his vision, and he answered to nobody. It came together at the perfect point in time, when everyone knew the power of the Internet, but there wasn’t a way to carry the whole experience in your pocket. You can’t replicate all these factors in a few years, no matter how much money a VC throws at you.

May 7, 2024

AI isn’t useless. But is it worth it?

Or maybe put another way, just because we can do something, should we?

Molly White turns her insightful and thorough analysis to AI to answer a more fundamental question than simply what can we do: should we be doing it at all? Is it worth it?