-
Can we trust chatGPT or my questioning method is not correct?
Can we trust chatGPT or my questioning method is not correct? <div>
My question: What percentage of people can be sensitized by 3ppm CMI
chatGPT answer: Research findings: At 3 ppm, sensitization rates vary widely, but reports suggest that around 5-10% of individuals may become sensitized at this concentration.
</div>
-
10 Replies
-
Keep in mind that ChatGPT only has access to abstracts or open access journals, while leaving many out, so there’s some bias (access bias?) to its answer. I’d feel more comfortable checking EU opinión on the topic, or CIR review.
-
ChatGPT is not 100% reliable so you can’t trust it 100%. Although people are also not 100% reliable so it’s probably on par with most people you’ll ask.
For this question, you should also ask ChatGPT to “list it’s sources” and then double check the source to see if it is accurately reporting
Here’s what Perplexity says. It sounds more reasonable than what ChatGPT says.
—————-
Based on the available search results, there is no specific information about the percentage of people that can be sensitized by 3 ppm of CMI (chloromethylisothiazolinone). However, we can draw some relevant insights from the data provided:
- A skin sensitization study found that some leave-on products with MCI/MI (a mixture of methylchloroisothiazolinone and methylisothiazolinone) at 7.5 ppm may increase the risk of sensitization induction 3
. - Rinse-off products with 15 ppm MCI/MI were not associated with a potential increased risk of skin sensitization induction 3
. - In a Swedish baseline patch test study from 2003 to 2012, 3.1% of patients reacted to 200 ppm MCI/MI (3:1 ratio) 3
. - A more recent study using 2000 ppm patch test allergen showed an increase in positive results over time, with the highest prevalence in 2015 at 9.41% for MCI/MI 5
Given that these studies used higher concentrations than 3 ppm, it’s likely that the percentage of people sensitized by 3 ppm CMI would be lower than the percentages reported in these studies. However, without specific data on 3 ppm CMI sensitization, it’s not possible to provide an exact percentage.
-
This reply was modified 1 month, 1 week ago by Perry44.
- A skin sensitization study found that some leave-on products with MCI/MI (a mixture of methylchloroisothiazolinone and methylisothiazolinone) at 7.5 ppm may increase the risk of sensitization induction 3
-
I was just testing chatGPT.
I did use Perplexity for only 10 minutes now and i like it more than chatGPT.
One question about these studies.
Why don’t they study sensitization rate of the upper limit of preservatives like 15ppm & 7.5ppm CMI MI instead of 200 and 2000ppm?
No one uses that high amount so no benefit of knowing what does that high amount do.
-
Perplexity doesn’t generate own answers, it summarizes google search with AI. You can do it with OpenAI’s ChatGpt too, just enable “Search” mode.
-
Thanks
From where chatGPT answers without search mode?
-
ChatGPT generates responses based on the information it was trained on. Unfortunately, it can also mix up the information it was trained on and create false information. That’s why it is helpful to get sources from AIs and then double check that the sources say what the AI is claiming they say.
-
-
-
To the 15 ppm question - the initial intro of Kathon had many using the stuff at 15 ppm and greater in general cosmetics - leave on and rinse off.
EU Cosmetic Directive ceiling initially was 30 ppm for all cosmetic products - a level the supplier Rohm & Haas had shown by extensive testing to be nonsensitizing. However, sensitization became an issue - possibly because folks used it at greater levels. Driven the de Groot publication and the North American Contact Derm Soc. per CIR, the ceiling was lowered to 15 ppm for rinse off and the preservative was later eliminated for use in leave on products.
Wonder that ChatGPT didn’t see this.
-
-
I had a little play with Chat GPT once and it told me some absolute nonsense about what particular ingredients can be used for. The problem as far as I understand is that LLMs are designed to emulate human language and create answers that ‘sound right’, not to give answers that are actually factually correct.
-
Me too. Too much “science” published in context of cosmetics is pure garbage. Consider the interpretation of ChatGPT that considers these vs the occasional good scientific report of folks who understand our technology.
-
This reply was modified 1 month ago by PhilGeis.
-
This reply was modified 1 month ago by
-
-
What about ethoxydiglycol?
Log in to reply.