Be prolific, not perfect
The speed at which AI is improving in different fields really hit me in a visceral way over the last couple of weeks. In photography, there was the first viral AI image mistaken for a photo, Pope in a Puffer; the AI ‘portrait’ that would have won a Sony photography prize if the creator hadn’t revealed its provenance, and the first AI ‘photo’ to fool me in the wild. I tapped on the latter in my Discover feed so that I could check out what seemed to be an excellent doc-style portrait of a young woman in a diner. It took reading the caption to recognise that it wasn’t made with a camera and that the woman and the diner don’t exist.
I’ve understood for a while that AI ‘photography’ would cannibalise the product, ecomm, stock, and maybe fashion spaces. But I thought genres like documentary or street photography, where authenticity is a cornerstone of their appeal, would be safe for a while. Perhaps that’s true philosophically, but soon it’s going to be nearly impossible to tell which bodies of work are real and which are synthetic. The ability to create consistent characters in believable environments is improving week on week.
Regarding music, I’d been predicting to friends that we would see an AI-generated hit within 1–3 years, maybe 5, but now I’m radically revising that down. I heard the pseudo-Drake x The Weeknd track Heart on my Sleeve last week. Not only does it sound like them, it’s also a solid commercial hip-hop track. I’d bet that it would be indistinguishable from tracks from the usual suspects in Spotify’s ‘Made For’ playlist. Instead of my previous ‘hmmm, cool trick’ response to most AI demos, hearing this was a definite oh shit moment.
I’ve known intellectually that AI would overtake all the drudge work, and then eventually all the aspirational occupations, but this is the first time I felt it emotionally. I want to get more hands-on experience with the key tools to gain a tacit understanding of what’s possible. Consuming breathless articles and podcasts about singularities, rapid takeoff scenarios, and the end of work, won’t actually ground me in their real-world applications or the best way to leverage them for their creative possibilities. I think that we’re in for an insane few years…
Two great and new to me places this week:
Recent discoveries added to the Want to Go list:
Saw two shows on Saturday:
I’ve been fiddling with Raycast all week. As a long-time Alfred user, it’s been interesting to switch out my app launcher. Alfred still feels snappier, and I prefer its approach to snippets (having a dedicated shortcut to pull them up immediately), but otherwise, I’m really enjoying the change. I like that much of Raycast’s standard functionality e.g. natural language input for calculations and currency conversions, required slightly janky workflows in Alfred. I’m signed up for the Raycast AI beta, so we’ll see how I feel when/if it gives me automation superpowers. I have a lifetime license for Alfred, so there’s no pressure to make a choice yet. I’m going to keep dialling in my Raycast setup and will see what sticks.
I’ve been experimenting with Narrative Select for culling the crap from big shoots. It’s good at ditching the obviously bad pictures, those with funny expressions or blinks for example, but it’s hopeless at pushing good pictures to the fore. Unfortunately, sometimes the best pictures also get pulled into the ‘rejects’ stream if they’re very dynamic — people laughing or moving and so on. It would be great to have help with the initial cull before I finesse the edit and select hero images, particularly while on a tight deadline, but it’s nowhere near ready for prime time. Its best feature has nothing to do with AI — I love how snappy it feels when flicking between pictures. It’s the main reason that I often use it for my first run through the images, before shipping the one stars to (slow and bloated) Lightroom CC. See Craig Mod on Fast Software, the Best Software for a refined take on the pleasures of speed.
I’m rawing near the end of The Creative Act by Rick Rubin. I’m going to do a write-up of my notes once I wrap it. About 40 pages left. I’m still reading The Path of Aliveness by Christian Dillo too…
Read a couple of good articles over the last few days:
…traditional architecture has always tended to be structurally dishonest. So if this is what makes contemporary traditional architecture pastiche, then most traditional architecture has been pastiche since the faux timbering of the Parthenon. Contemporary traditional architects have most of the great builders of our history as their companions in guilt.
The modernist critic thus has two alternatives: either to concede that neither the modern nor the premodern traditional architect is a pastiche artist, or to claim that both are, and hence that Greek, Roman, Renaissance, Baroque architecture – and potentially many others – should all be condemned as fake. The latter option is consistent but hard to take seriously. It is difficult to resist the conclusion that, however important an understanding of structure may be in our appreciation of certain buildings, structural honesty is no more necessary to good architecture than originality is.