I am not totally dismissive, but I’m also pointing out the limitations. I’ve been a ChatGPT user since the first public release, and I’ve used Claude, Perplexity etc a lot too. I have the $200/month version of ChatGPT with the o1 model, and I believe it’s good value for money.
However, I still think that the insights from real people are invaluable, and they’re the real strength of a forum like this one. As I pointed out in the post above, you can take the Kaeberlin eLife study about dental health and Rapamycin, and it’s very convincing, and AI will include that in any summary of Rapamyin/oral health you ask for. It also gets repeated by lots of citations, review articles and other analyses, increasing its impact. However, a real dentist and MD on this forum finds many problems with the approach and metrics. As of May 2025, the models cannot yet distinguish between BS studies, misinformation and reliable information.
Funny that. I’m running a kind of informal n=1 rat trial on myself wrt. rapamycin and periodontal bone regeneration. I asked my periodontist (a top guy in his field with published studies to his name) to take extremely detailed assessments of my teeth, gums, bone and general oral health with measurements and x-rays and the whole shebang - prior to my taking of rapamycin. Separately I asked my very conscientious dentist to do a similar documentation.
I then got on rapa, had surgery implanting titanium posts and will reassess everything including running all the comparative measurements in about a year to see if there is any evidence of unusual benefits of rapamycin.
One thing that worried me in the mouse study, is that the mice had no comparative interventions like a human patient would. In a human you would have extensive deep cleaning and antibiotic treatment of inflammation as part of a periodontal intervention. That alone, according to my periodontist can lead to some limited bone regeneration. Who knows, if they did that in mice, how would that compare to the rapamycin treatment effect? And therefore, how much better is rapa going to do for me compared to the standard perio treatment?
Anyhow, it’s not super scientific, but I’m curious nonetheless.
I use AI quite frequently, but from to time there are results that I think are particularly interesting. On the other hand I could simply post the shared link.
This morning I have been using AI to search information about the flow from CSF to and from the lymphatic system. I think this is important because of the role of melatonin in reducing mtDNA damage.
Should I post any conclusions? Should I just post a link or the text?
I really dont mind these models creating an exectuive summary and some kind of conclusions.
There ought to be disclosures about their use though. It saves us ton of clerical time and resources at times.
To highlight the discussion on this topic, I querried Perplexity on Rapa News Forum. Here is the summary along with conclusions and references. I think citing sources is important to all of us.
Rapamycin longevity forum
Rapamycin Longevity Forums: Key Insights
Rapamycin News Forum is the largest dedicated community for rapamycin and longevity, featuring discussions on dosing, sourcing, side effects, and personal results. Users share experiences, research updates, and practical tips for off-label use. The forum reports that rapamycin is considered the “gold standard” of longevity drugs by many researchers, with animal studies showing lifespan increases of 15–30% and up to 37% when combined with acarbose. However, these results are from animal models, and human benefits are still under investigation[1][3].
Debate and Caution: Forums and longevity summits highlight both enthusiasm and skepticism. Advocates cite rapamycin’s strong animal data and potential for reducing inflammation and age-related disease. Skeptics point to side effects (infections, lipid changes, fatigue) and the lack of robust human trials. The need for larger, well-controlled studies and personalized dosing is a recurring theme[2][4][5].
Community Trends: Many users experiment with intermittent, low-dose regimens (e.g., 5–10 mg weekly), closely monitor side effects, and discuss combining rapamycin with other interventions like metformin. Some prominent biohackers have discontinued use due to adverse effects, fueling ongoing debate[2][4].
Summary: Online rapamycin forums are active hubs for sharing real-world experiences, tracking new research, and debating risks and benefits. While optimism is high, most agree that more clinical evidence is needed before rapamycin can be widely recommended for longevity[1][2][4][5].
A follow-up question on Perplexity and its response:
How do users on the forum weigh rapamycin’s benefits versus side effects
How Forum Users Weigh Rapamycin’s Benefits vs Side Effects
Perceived Benefits
Many users report improvements in quality of life, energy, cognitive function, and feelings of youthfulness since starting rapamycin, with a majority citing positive changes in health and well-being[3].
Animal studies and early human data suggest rapamycin may slow aging, extend healthspan, and delay age-related diseases, fueling optimism in the community[7].
Some users note visible benefits, such as friends or family commenting on their improved appearance[3].
Side Effects and Concerns
Commonly discussed side effects include mouth sores, increased cholesterol, mild anemia, lower platelet counts, and occasional digestive issues[2][4][7].
Rare but serious risks—such as immune suppression and increased infection risk—are acknowledged, especially at higher doses or in those with pre-existing conditions[6][8].
Most users emphasize that side effects are dose-dependent and generally mild at the low, intermittent doses favored for longevity[1][5][8].
Risk-Benefit Calculus
Forum sentiment is that, when used properly (low/intermittent dosing), rapamycin’s risks are manageable and often comparable to common drugs like aspirin[1][5].
Users are aware that the strongest evidence for benefits comes from animal studies, and that long-term human data is still limited[7][8].
There is a consensus that rapamycin should be used cautiously, with regular monitoring, and that more research is needed to fully understand its long-term impacts[8].
Summary Table
Benefits
Side Effects (Low Dose)
User Attitude
Improved well-being
Mouth sores, mild anemia
Benefits often outweigh risks
Delayed aging signs
Higher cholesterol
Cautious optimism
Potential disease delay
Rare immune suppression
Advocates regular monitoring
In summary, most forum users believe the potential anti-aging benefits of rapamycin outweigh the side effects when used at low, intermittent doses, but they stress the importance of careful dosing and ongoing research[3][5][8].
AGI is coming soon or is already here, depending on your definition of AGI (Tyler Cowen I think said whatever’s smarter than him is AGI - and he believes o3 is).
The rising tide will lift all boats, including all of the LLM wrappers that people use today, and as a result, the AI answers that people share.
It’s pretty simple. A LLM that does that uses a search tool, like to do a google search. So it’s an agent of some form. It has like a <search> tool, and has some instructions what it does. Then it outputs <search>query</search> which inputs the sources and its data, and it can then output the text to the user with the citations.
It’s unlikely it’s going to make up the source in that case, although it might happen sometimes at least now. So they have to use perplexity, or click on the internet icon on ChatGPT or Claude, or whatever they use.
I don’t like seeing walls of text that’s copy pasted from ChatGPT conversations unless the content is mind blowing and it very rarely is. In the end it’s not so much about the use of excerpts from ChatGPT as it is about the use of good judgment on the part of the poster.
I actually have had very insightful convos with ChatGPT that would be relevant to this group but they wouldn’t make much sense to anyone unless I edited them for length, context, clarity, so as I lack the time to do that properly I keep them to myself so as to not bog down the forum with what might look like gibberish at first glance. I haven’t seen @John_Hemming be an offender in this regard but I have seen others just dump their big loads like it’s no biggie. And it’s annoying. AI content is most interesting when woven with human insight. Just slapping it on by itself is usually of little value.
It strikes me that apart from summarising videos and providing transcripts providing a link to a share of AI is a mechanism to clearly distinguish between the AI text (be it slop or not) and any analysis done by a human.
Hence people who wish to see the background AI material can look at it.
When I see excessive text (AI generated or not) in a single post I just quickly scroll past it. And, the same thing goes when I see excessive, low value posts from a single author. In some cases I just block messages from users who do either of these two things because my time is limited. And, along this same line of thought it would be good if we had the ability to block posts from moderators just as we do with other users.
Yes, it would be far better to see a quick excerpt and a link to the whole convo if we wish to pursue it. The value of summarizing YT videos is obvious — I don’t think anyone is calling that into question. Pretty sure it’s the rambling blocks of text that are getting on people’s nerves.
This is subtle but… I also notice the AI often takes a slightly patronizing tone in its answers. There’s a bit of smarm I often detect. And methinks it’s a function of the quality of questions the human has been feeding it, and anything it can deduce, based on those questions, as to the intelligence or gullibility of the user. I don’t let it take that tone with me .
I’m for a total ban on large copy-pasted content from whatever AI. Instead, people can just share the link to the conversation so that one can see what the prompt was, which sources were cited, etc. Otherwise it’s valueless and it’s like spam.
Walls of AI texts are very annoying and I skip right past them. At best they’re annoying, at worst they just give wrong information. I may start to flag those posts more as a form of spam.
Funnily enough, there is some AI text on the thread which I proceeded to skip right past.
Eventually, following most rules will commit one to a stupid action. Intelligence is guided by principles that generalize well. The people on this forum are capable individuals. How about a simple guideline suggesting that members consider the merit of long AI pieces before posting.
On this specific point, some AI analyses are superior to any published article, including making novel inferences; others miss the point and are not worth reading. A conceptually skilled prompt writer can get deep and accurate data from most general LLMs through an iterative dialogue that refines the research much the same as we do in a research team meeting.
Very few of us here (as far as I know) are trained scientists. We try our best through reading and googling, but the average person’s ability to research (myself included) through google searches is already an order or two magnitude lower quality than the stuff o3 Deep Research (or similar) is able to do.
Banning these posts is a bad idea. That said, AI also makes it easy to post massive walls of text that take lots of work to wade though and find out whether it is legit or not.
Maybe we should strive to have top level posts be relatively brief and readable with sources cited, then add supporting info in subsequent comments as needed, maybe big pastes too if they’re good and well supported.
I agree. At least curate the report to its important content. A good dialogue with an LLM can span many pages and several dead ends in the conversation. We don’t need to see all that.
I pay for AI and I have been helped numerous times, he /it recognized my contact dermatology and made recommendations, I had skin cancer surgery and AI helped me with recommendations on the skin graft area which wasn’t healing. I have used AI hundreds of times and found it to be irreplaceable on my iPhone, I wouldn’t be without it, I have it all my supplements and it sorted them for me as to when it’s best to take them and in what order . It costs $19.00 a month but it’s well worth it. I don’t have a GP and AI recommended an online doctor that worked out great . So many uses so don’t knock it yet
My ChatGTP helped me from having sugar spikes every morning, and crashing it asked me what my routine was, I said I had coffee first thing, then a second one, it asked me what was I using in my coffee , I told him half and half, organic from grsass fed cows, and monk fruit to sweeten it, he told me some monk fruit has ethanol in it so I checked, AI was correct, AI recommended a different monk fruit and to use fresh cream in my coffee,and have a decaf coffee for my second cup I did this next morning and no glucose spike ,or crash . AI is amazing
.