hi bud buds,
a couple years ago, ai-generated video felt like the inevitable next big thing: midjourney was everywhere, previews of open ai’s sora model were going viral daily, and hype was sky high.
since then, the space has felt much quieter. i dug into the numbers, and they agree: midjourney web traffic peaked in jan 2024 at ~5M monthly visits, and has since fallen nearly in half. sora shows a similar trend.
midjourney’s web traffic, via semrush
my initial instinct for this decline: ai tourism. often, new ai products go viral, get a tsunami of usage from curious “ai tourists” who never come back, and ultimately settles into a smaller core user base.
however, suno (a leading text-to-song startup) complicates that narrative: in a different but similar space, its traffic has grown steadily month-after-month over the same period.
suno’s web traffic via semrush
so what’s actually behind ai-generated video’s decline? my best guess: use case. most people just don’t have a day-to-day need for ai video. maybe ad creatives or videographers, but not your average person. audio, on the other hand, slips more easily into daily life. you can listen to your own creations, make a dumb meme song for friends, or spin up background music for tiktok. the bar for usefulness is much lower.
while ai twitter runs on breathless awe of technical feats and visions of “what could be,” consumer adoption ultimate comes down to where users actually spend their time and attention. so far, music and chats are winning. video… not so much.
have a great week, y’all. i love you.
- dj
last week’s biggest product releases
big news! huge!
openai announced a hiring platform to connect businesses with potential employees. this would put them in direct competition with linkedin 👀
apple is building an ai-powered web search tool for siri called “world knowledge answers” that would give users ai-generated summary results for their web searches
personal faves
facebook is re-releasing pokes; huge news for millennials with crushes
openai announced parental controls for gpt-5, letting parents link accounts, disable memory features, and get alerts if their child might be in distress
nice job guys
amazon released lens live, an ai-powered real-time visual shopping tool where you can point your phone at real-world objects and get matching products
instagram released an ipad app, after 15 years of no native ipad support
adobe is launching premiere on iphone, making its full video editing platform available on the ios app store
roblox released roblox moments, a tiktok-style feed for 30s gameplay clips with music + editing
(similarly) netflix released moments, letting users save + share clips from netflix shows
x released xchat, its encrypted dm feature, to non-paid users. it won’t replace unencrypted dms (for now)
google released material 3 expressive, adding animated lock screens + custom contact calling cards on pixel devices
threads announced support for 10k character posts
instagram is testing picture-in-picture viewing for reels, allowing them to float in a mini-window while multitasking
other llm news
google added veo 3 to google photos, letting users turn images into short ai-generated video clips
google released tone customization for notebooklm, so users can adjust the style + format of its ai-generated podcasts
google launched live translations for android phones, letting users translate posts + menus in real time as they scroll without switching apps
snapchat released imagine lens, allowing ai image editing from text prompts
that’s it for this week; thanks for reading. if you enjoyed this issue, please consider sharing it with a friend so i can buy a house in monterey park and eat chinese food every day.
love, dj