Dr Michelle Wong - Lab Muffin Beauty Science

Here’s why AI chatbots (Claude, ChatGPT, Gemini etc) are unreliable for science content, and terrible as “just a starting point” for science communicators. Unfortunately, this probably isn’t something AI chatbots get much better at… although sloppers will get better at hiding it!

Thanks so much to Dr @jordanbharrod for helping me with this 😊 This is adapted from my YT video on AI slop scicomm, which she also helped me with! Her work is an excellent resource for understanding AI hype. Will post a proper article version of this soon!

I've probably been the top sunscreen mythbuster online for the last decade, so I have thoughts! It's super cool that sunscreen misinfo is being formally studied! Big props to the authors for recognising this problem. But there's misinfo in the paper, and IMO the main claims aren't well-supported…

I recently spoke at the Science Misinfo Symposium - later, some social media researchers told me they weren't aware of some conventions mentioned in my talk. IMO low-hanging fruit for getting better data about social media would be to get input from content creators - many would be happy to help!

I've probably been the top sunscreen mythbuster online for the last decade, so I have thoughts! It's super cool that sunscreen misinfo is being formally studied! Big props to the authors for recognising this problem. But there's misinfo in the paper, and IMO the main claims aren't well-supported…

Perceived institutional endorsement is IME very important for the spread and impact of sunscreen misinfo - authority bias is powerful! I've seen many people cite intros of peer-reviewed papers as "gold-standard science". "Minerals reflect" is the origin of many sticky myths repeated by derms.

I've probably been the top sunscreen mythbuster online for the last decade, so I have thoughts! It's super cool that sunscreen misinfo is being formally studied! Big props to the authors for recognising this problem. But there's misinfo in the paper, and IMO the main claims aren't well-supported…

So while IMO sunscreen misinformation probably does get more engagement, I don't think this data is good evidence for it. I think it's just finding that clearly disclosed paid advertising gets less engagement than the organic content people came to the platform to see. Not exactly surprising!

I've probably been the top sunscreen mythbuster online for the last decade, so I have thoughts! It's super cool that sunscreen misinfo is being formally studied! Big props to the authors for recognising this problem. But there's misinfo in the paper, and IMO the main claims aren't well-supported…

I also did a quick pass over the rest to mark the ones I thought were likely boosted. Boosted views were likely 38.5-49.5% of the total views used in the analysis - a decent confounder?

I've probably been the top sunscreen mythbuster online for the last decade, so I have thoughts! It's super cool that sunscreen misinfo is being formally studied! Big props to the authors for recognising this problem. But there's misinfo in the paper, and IMO the main claims aren't well-supported…

TikTok doesn't seem to think so either - they count boosted views as less when you sort by "popular"... (Note that this means TikTok has the data! They could share it with researchers!)

I've probably been the top sunscreen mythbuster online for the last decade, so I have thoughts! It's super cool that sunscreen misinfo is being formally studied! Big props to the authors for recognising this problem. But there's misinfo in the paper, and IMO the main claims aren't well-supported…

My normal sponsored post view count at the time was ~23.5K. This means roughly 99.7% of the views on my boosted videos were "inorganic" - paid to be shown to more people, and clearly labelled that way!

I've probably been the top sunscreen mythbuster online for the last decade, so I have thoughts! It's super cool that sunscreen misinfo is being formally studied! Big props to the authors for recognising this problem. But there's misinfo in the paper, and IMO the main claims aren't well-supported…

"Boosted" ad views are not the same as regular views. * People swipe away clearly labelled ads quickly without registering them. * Boosted ads aren't matched to users via the platform's algorithm, so they're inherently less likely to engage the user.

I've probably been the top sunscreen mythbuster online for the last decade, so I have thoughts! It's super cool that sunscreen misinfo is being formally studied! Big props to the authors for recognising this problem. But there's misinfo in the paper, and IMO the main claims aren't well-supported…

Brands often boost videos they paid creators to make (ads), plus their own posts. I didn't realise this was an issue until I saw the SI… Of the top 10 videos in their dataset: 6 = clearly disclosed ads 2 = posts from brands 1 = likely undisclosed ad Only 1 was likely organic (regular) content.

I've probably been the top sunscreen mythbuster online for the last decade, so I have thoughts! It's super cool that sunscreen misinfo is being formally studied! Big props to the authors for recognising this problem. But there's misinfo in the paper, and IMO the main claims aren't well-supported…

I think the inaccurate estimate comes from issues with the methodology: ISSUE 1 Videos were found using sunscreen-related hashtags. Hashtags have declined in use since ~2022, especially on TikTok - the algo is very good, so hashtags probably don't reflect what people are seeing.

I've probably been the top sunscreen mythbuster online for the last decade, so I have thoughts! It's super cool that sunscreen misinfo is being formally studied! Big props to the authors for recognising this problem. But there's misinfo in the paper, and IMO the main claims aren't well-supported…

CLAIM 1 "Overall, the dominant narrative on TikTok promotes sunscreen use" "Our findings indicate that [sunscreen misinformation] constitutes a small portion of the most viewed videos" They estimate only 1.5% of highly viewed sunscreen TikToks state it causes harm.

A lot of chemical #sunscreen fear is from animal/in vitro studies taken out of context, without comparing exposure in those studies to the actual amounts we're exposed to from sunscreens. Safety limits have massive buffers (100-fold margin of safety, from eg. no observed adverse effect level).

It's also important to consider the downsides of the alternatives - mineral sunscreens have a lot of drawbacks, very few people end up being able to use them regularly (if you genuinely prefer mineral sunscreens that's fine, just watch out for separation and adequate application!)

My time to... shine? 🫠 It's a bizarre experience to have this much attention on my very niche area of expertise.

Is your scalp itchy and nothing seems to really work? I got you

Seeing a lot of people recommending sunscreen mists... unfortunately they're really not that effective for applying to your face, sorry! See the cloud lingering in the air? That's sunscreen NOT going on your skin - my experiments estimated that 80-95% of the sunscreen doesn't land.