Hacker News Reader

Comment by sangnoir | Hacker News Reader

> [...] the founders were not some random people who didn't understand US culture...This is important to the article's thesis.

This is the opposite of the article's thesis. Musical.ly hit a wall. Only after it was purchased by Bytedance, rebranded as TikTok, and Bytedance poured their magical, cultural-agnostic AI sauce (and ad dollars) did TikTok truly take off. By all indications, this AI magic was wholly created in China and retrofitted onto TikTok.

The article's thesis is that AI doesn't need to understand the culture when it can understand the person at an individual level. The closing anecdote emphasizes the point: a Chinese startup that has a number 1 news app in India and not a single person in the company can read any of the content in their app (Hindi)?! How is that for not understanding a culture?

View on HN

Musical.ly hit a wall. Only after it was purchased by Bytedance, rebranded as TikTok, and Bytedance poured their magical, cultural-agnostic AI sauce (and ad dollars) did TikTok truly take off. By all indications, this AI magic was wholly created in China and retrofitted onto TikTok.

Music.ly blew up in 2015 and was in the top app charts for a year before Bytedance bought them. They were at 133 million registered users and 40 million MAU in 2016. What was the wall they hit?

https://digital.hbs.edu/platform-digit/wp-content/uploads/si...

They "hit a wall" in the sense that they couldn't keep up the hockey stick style growth that venture capitalists have come to expect.

Maybe that's true, but in that case it was just Bytedance as a financial savior, not technical or product savior. Could have been Facebook, Google, or anyone with deep pockets.

As an AI-focused human, I'm pretty skeptical of the idea that ByteDance' AI made the difference here - not because I dispute that AI can pierce cultural boundaries, but because doing so is a purchasable commodity for a scaled social platform. The magic sauce here was getting the flywheel turning to have users invent content templates native to the medium. Ad dollars, as you mention, and some smart decisions about biases (maybe, as another commenter says, the creative positivity that survives CN censorship).

Right, I meant that this fact undermines the article's thesis. It's important to consider when evaluating what happened with TikTok.

I think the general point is right (culture can be abstracted), but I am skeptical that ML/AI alone would have been sufficient.

Reply on news.ycombinator.com