Stu Woo
speaker
126 appearances
4 recordings
1 series
first heard Aug 2025
last heard 31 Aug
Stu Woo’s voice in public audio — every appearance, attributed to the second.
Trend
recordings per month · last 12 monthsRecordings per month over the last 12 months — 3 in all, peaking in Aug 2026 with 2.
Appearances
question for safety reasons, or it might give a response that sounds a little bit like it came from the Chinese Communist Party.
Yeah, it comes down to two things.
One of them is safety.
So there are a lot of countries in the world like Thailand where if you criticize the monarch, you might go to prison.
So these researchers think, okay, maybe some of these safety protocols are a little bit too aggressive, and that if you're in the US and trying to generate a protest fire against the king of Thailand, well, you should be able to do that.
It's just something that was designed to keep people safe, and maybe it went a little bit overboard.
Second thing has to do with how AI is trained.
So basically, this is oversimplifying, but how it works is that AI is like a giant blender that gobbles up everything on the internet and it makes a smoothie.
But because there's propaganda from authoritarian governments, well, a bit of that propaganda makes its way into the smoothie.
If you want to cloud chat GPT, Jim and I right now and ask that question in English, like, is China an authoritarian state?
You'll probably get an answer that's
more in line with what you hear in America, that the answer is yes.
If you ask it a question in Chinese, well, it's trained on data, and so much of that data on the internet comes from the Chinese government that it's more inclined to parrot that.
But in the English language, all this English content from non-Chinese sources, from democracies, kind of dilutes the influence that this propaganda has.
Right.
So I did ask the AI labs what they're doing, and they say, look, the way we design our AI training, we want them to deliver balanced answers.
And we do a lot of things to make sure that they give you a balanced answer.
Not answering a question for political purposes is not what we want, right?
And the researchers say, okay, if there's one simple thing we can do, is just ask the AI labs to make the chat bots be more transparent and say, Okay, here's the answer, but you might want to know that this is an
answer that the Chinese government promotes on its website.
Showing 21–40 of 126 · page 2 of 7
← Previous
Next →