The Sloppification of Peptides(henryaj.substack.com) |
The Sloppification of Peptides(henryaj.substack.com) |
Training corpuses (corpii?) have always had lies, propaganda, derangement, you name it.
In truth, I don't think that's quite how it works, but I wouldn't be surprised if teams who scrape data for training apply weights in favor of data that is known to be generated prior to the GPT-3 era.
Btw, learning from your own generated data isn't necessarily bad, if you have an external source of validation. See how AlphaGo learned from self play for a simple example.
Source: look at how good Chinese models are.
I'd say at least 99% of all new youtube videos are AI generated video, whether taking snippets from other videos and voiceovering them, or just making entire fake personalities up from scratch. That number will raise to 99.99% soon.
There is no veracity to these videos, and within a year I'd say no one will ever be able to detect AI or not. This is exasperated by the fact that normal, real productions are being AI edited, and metadata or other non-visible means to detect AI or not will be meaningless.
Bluntly put, most people currently get their 'news' from Tiktok and Youtube. When I discuss current events with random, not-tech people, they all "believe" that if it's video, and on youtube or tiktok? It's real.
I cannot express how much the average Joe thinks, in their mind, "Something published on the Internet" is high in trust. I'm verbose, I talk to a lot of people, and there is an inverse relationship to computing skill, and Internet trust.
And using a computer to "do things" is not "computing skill", any more than driving makes one a mechanic, or ensures you understand a car engine.
I guess my point is, I suspect that the continued degradation of reporting, and news, conjoined with alternate sources being 'news sites', will continue to the point that "reliable source" is non-existent.
For example, imagine if every single post from a phone, such as "I see troops in City X' are seen as almost certainly faked. Or even, "I don't see troops in City X'. And if you think of it, certain sectors of society preferred the curated, more controlled news of the past. There may be attempts to purposefully destroy the credibility and believability of every and all independents on any platform.
Create a personality, have it trusted for months or years, and then show that its data was false. Doing this is easy with the tools today, doing it by creating one million such personalities with the same goal, is just as easy.
The end of truth for all is here now.
There is no point of trust now on the Internet. The cut off isn't just for AI, and training.
The cut off is for humans believing anything.
Look at what companies like Telus or ScaleAi are doing. There are probably tens of thousands contractors at the moment sitting carefully tuning LLMs with quality, vetted, question answers.
It is even simpler because AI companies don't need to be that precise, just a classifier that can tell roughly what is slop and what is not, missing some of the non-slop is no big deal, and having having a bit of slop get through is not going to change things much. LLMs are relatively robust, otherwise considering the amount of bullshit we find online, they would be completely useless.
Yes, <bad code, fake pictures, security vulnerabilities, malicious actors> existed before LLMs. No one is saying otherwise.
The blog post rightly points out we should get good at spotting all this but this is not the scary part, the scary part is that the audience for these ecosystems is not even humans, it's the big AI players like Gemini and ChatGPT, and those are places that a lot of people are going to get information and advice on things.
I use Gemini quite a lot and lately I've already noticed it's very quick to give me a broad spectrum vibe on some issue or topic and not good at all at backing it up with reputable sources. So I think this is potentially a very ominous development of the entire web itself becomes a giant poisoned cauldron of untrustworthy AI maliciously influencing "trustworthy" AIs.
Update: i fed this comment into Gemini and asked "what is Google doing about this?" and it produced a very long, detailed answer, leading off with tagging this as the so-called "internet ouroboros" problem. For those who don't feel like doing the same, I can distill it's answer as follows: "Yo. Chill. We got this, bro. Lol" so we're good!
No, they don't, it's a very Silicon Valley specific thing and I never heard about them here in France. People need to stop thinking their experience generalizes by default
It is legal to sell research only peptides that are not FDA approved, but as soon you talk about efficacy and human use, you also become a target for the FDA warning letters. [1] [2] [3].
Anyone selling these peptides who are indexed by search engines face increase scrutiny immediately, especially if you're going to try to sell the patented ones, because then you're a Target of not just the FDA but also Eli Lilly... peptides are big business but it's big risk if you do it wrong.... Eli Lily wants people spending $1,000 per month to use these peptides.
I think the trick about selling peptides to people is to probably do it word of mouth, no indexing by google, no ads, double gated website with invite only structure.
[0] https://www.npr.org/2026/08/13/g-s1-138594/eli-lilly-files-6...
[1] https://www.fda.gov/inspections-compliance-enforcement-and-c...
[2] https://www.fda.gov/inspections-compliance-enforcement-and-c...
[3] https://www.raps.org/resource/fda-warning-letters-target-mar...
That being said, there’s usually a reason they were abandoned by the pharmaceutical industry; they were ineffective or had a worse side effect profile compared to other drugs already in the pipeline. Good example of this is the “Barbie drug” melanotan II which was abandoned in favor of bremelanotide for treating hypoactive sexual desire because the latter targeted fewer receptors and was less likely to cause hyperpigmentation and other side effects.
Ultimately, most people should probably just stick to pharmaceuticals that have been FDA approved.
>"The trained eye, weary of vibe coded designs, will see Claude Code’s aesthetic fingerprint all over this. "
yeah, everyone is re-doing old interfaces with LLMs.
>"They're hosted on AlexHost"
yeah, they're selling grey market pharma.
Now, I have no problems saying that the whole thing is a bot driven scam -- it probably is , grey market web is filled with scams -- but there was no evidence given; just an ad for a magic tool that detects AI... somehow.
...which makes me feel like an asshole for putting in the time looking for the anchor and hoping it wasn't an advertisement.
Try browsing through the offtopic forum [1] if you're not convinced. About half of the threads lead with a post which uses the phrase "food noise" (often "the food noise came back"), even in topics like "Pet photos thread" or "TV shows to binge while on the treadmill" [2] where that content is completely inappropriate.
Or, along the same lines: try reading the "welcome thread" [3]. No one actually introduces themselves; all of the replies immediately start treating the thread as if it's a discussion of some vague topic. "Admin" even comes back several times to "close the loop on my own question" and to return "with an update, since a thread like this is useless without one" [4].
In short: all the simulated user behavior on the forum is wildly unnatural. It doesn't stand up to any close examination.
[1]: https://www.compoundtalk.com/forums/offtopic/
[2]: https://www.compoundtalk.com/threads/t29681/, https://www.compoundtalk.com/threads/t31154/
[3]: https://www.compoundtalk.com/threads/t34774/
[4]: https://www.compoundtalk.com/threads/t34774-page-10/#post-47, https://www.compoundtalk.com/threads/t34774-page-20/#post-98
Sadly, my view on the AI literacy of people who gleefully use 'slop' stays steady... which is a shame. Cause we're all on the same side, in the end.
* Use pangram for catching AI slop
* Be good at spotting AI slop
* Beware of website built for AI crawlers only
* Do not buy stuff from China
* Do not consume peptide supplements
* Peptides are good but buy local
* Do not live in SF
https://www.abc.net.au/news/2026-08-05/unregulated-peptides-...
I shall consult the zoomers.
Maybe a US thing
They learn about it through Insta or Snapchat and take whatever is promoted by whatever big muscular guy.
haha, good one! ++1
While I doubt, actually, just for the reason that a) a boulangerie is much cheaper to open and that there are in general b) more people who need a boulangerie more necessarily than a gym?
:-))
I've never heard an actual human in person talking about them.
it's not a sv-only thing.
Not every aspect of american culture makes it overseas and much less does now than used to 20-30 years ago.
I think peptides (and medicating yourself out of a hole in general) are a pretty uniquely american thing.
Also, I'd add that gym bros definitely have heard of them, especially if they are familiar with steroids.
The GLP-1 drugs are probably the main cause of this. We have this amazing class of compounds that work really well, But are priced pretty high, at least considering lower UK wages.
So there is a big incentive to look at getting these much cheaper from China. You hear about this happening, and then you see the fact that these companies have big catalogues full of other peptides that are a crypto payment and a click away.
I think, if you wanted to find out about the topic you mention, you might first want to look for literature that would support such claims (even preprints on arxiv would at least be a start) before looking for actual products.
The real question is: Should you be putting novel, poorly researched and dubiously manufactured substances (which happen to be peptides) in your body via injection? To that I say, to each their own but I'll wait for the pharma industry to catch up. I'm not that bored.
on one hand: no, absolutely not, hell no.
on the other hand: these people are voluntarily doing human sperimentation. things like this are going to happen anyway so we might as well collect the data (samples, effect etc). it's not much different from the safe needles initiatives.
Life out of balance, indeed.
If you artificially inject the correct protein into your body to mimic a naturally producing one, it would do the job and in that way peptides works. However, most of peptides injected by people are hardly human tested for long term effects. Human body is remarkable at adjusting and maintaining a baseline and we don’t know what would be the unintended side effects of these peptides because the body would try to counterbalance it. At best it could be just a mild headache or stomachache and at worst it could be cancer
Anyone talking about "taking peptides" as a meaningful sentence is absolutely dismissible in the same way someone talking about "taking chemicals" is dismissible.
Injecting poorly-studied grey market peptides seems like a bad idea.
1) PT-141 is a peptide that releases sex hormone into the blood and literally makes you horny, and its primary use is for premenopausal women that have low sex drive.
2) PNC-27 is a peptide that supposedly attaches to the cell membrane of tumor cells and performs cell lysis (death) of the tumor cell, and it's supposedly affects many different kinds of tumor cells: breast, pancreatic, lung, colon, ovarian, cervical, and leukemia
3) Epithalon is a peptide that is noted for extending telomers on chromosomes, leading to longevity benefits.
4) MOTS-c is it peptide that in animal studies is shown to boost physical endurance and mimic molecular changes associated with training/exercise.
From what I've read there is some real risks associated with using them. Sounds a bit like the marketing of steroids in the 70s.
Unfortunately many peptides break down in the GI tract. However there is some reason to believe eating protein, like collagen in meat does produce an outcome in terms of muscle mass and skin condition. I guess because those breakdown products make their way through the blood/lymph system to where they're needed.
All those people huffing Whey powder after Gym? peptides! Just enjoy a low sugar yoghourt at lunchtime guys. same-same
Please tell me you're kidding.
50 years ago it was cocaine, 40 years ago there was crack, 30 years ago it was heroin, 20 years ago we had bath salts, 10 years ago it was gas station drugs and meth, 5 years ago it was Ivermectin, this year it's SF techies doing their own research into peptides and taking some. Taking the latest drug (or peptide in 2026) is its own benefit.
As with any substance, there can be use, and abuse. If you would have told people 30 years ago you use creatine, you would be stared at as if you inject steroids or growth hormone. Today it is in the supermarket, literally next to the protein bars. Throwing Ivermectin on the same pile as crack and heroin though.. :D
I am sure there are dozens (dozens!) of idiots who inject whatever they think it is for whatever reason, valid or not. Peptides is a craze, and it is also emerging here in the Netherlands, but as people have pointed out, lots of things are peptides, and lots of them seem to have significant health benefits, when used properly. Give it another 30 years... they will be next to the protein bars as well, once someone has figured out how to capture the profits.
I used to be on a biohacking/nootropics discord a decade ago. People were trying to treat their depression by ordering research chemicals from Russian labs or synthesizing them themselves. Huge difference between that and going to the GP and getting a Prozac prescription, with semi-regulated supplements being in the middle. That's analogous to ordering peptides from sketchy suppliers, getting a GLP-1 prescribed by a doctor, and getting a peptide facial from a medspa.
I would assume that an order of magnitude more people are getting prescription GLP-1s than are ordering other peptides from online suppliers and injecting them.
Telomeres are part of your anti-cancer mechanisms (they stop cell reproduction after say 60 DNA self-copies to help fight unbounded cancerous growth).
If chasing 'benefits', you'd want to be careful before fucking with that mechanism.
And, as you say, the gym performance / sports crowd.
The BBC and the Guardian have covered it fairly neutrally, but the likes of the Daily Mail and gossip mags have been both more frequent and more breathless.
:-)
For the vast majority of these chemicals, there's no evidence they do anything beneficial. Until there is, you're an idiot for taking them and our regulators are idiots for letting them be marketed to you as if there are benefits.
The entire conversation is goofy because once there's solid enough evidence of safety and efficacy, we just call it a "drug" like any other drug, but somehow a lot of the voluntary lab rats in this craze trust a gas station kiosk more than a billion dollar RCT.
TLDR I agree with you
What disease does snake oil cure exactly according to you?
It is these American counterparts that dissembled the truth by creating ineffective variations of the formulas hence the twisting of the term, "Snake Oil Salesmen."
NRP Article on History of Snake oil Salesmen https://www.npr.org/sections/codeswitch/2013/08/26/215761377...
The drugs (coincidentally of shape "peptide") that have been through clinical trials are obviously measured against placebo, and they (like the approved GLP-related set of drugs) are outrageously effective beyond placebo. And have very little downsides compared to the downsides of being obese, which are far more immense than people appreciate.
Narrowing down areas into "Anecdotally many people say it works, we should check" does have some value, even if it does just wind up debunking the claims
There are far more promising development targets than there is development resources. Self-reports from a bunch of the most drug-addled people in the world are not going to make it to the next-most-viable target.
We're amused by your antics but no one wants to emulate American culture anymore.
if I asked 20 of my friends about peptides I'm probably the only one that subscribes to that american leaning, peptide-spammed subreddit.
However given that in English we don't care eg about Latin cases, I don't think we should care all that much about Latin plurals. So corpuses is fine by me.
"Corpi" gets used as a hypercorrection, but it's deprecated in favor of the two previous.
"Corpii", as suggested above, is right out.
I often use corpora, because it's what's common enough. And I sometimes use nonsense plurals like octopodes just to be silly.
The reason I replied is because the OP appeared to be interested in the most accepted or used plurals for this word.
As you say, insulin is a fully fledged protein, albeit a small one, and i wouldn't call it a peptide myself. But then, insulin is 51 residues, whereas tirzepatide (Mounjaro) is 39; is that such a big difference? FWIW i think a lot of other "peptides" are even smaller, eg CJC-1295 is about 30 residues.
(Yes, eating healthy and exercising would probably help most people, but many remain)
And for the remainder: go to your physician to get to a state where you can do healthy diet and exercise.
For the small part that remains is highly suggest getting an actual physician to give you treatment options, instead of self medicating with sketchy Chinese chemicals.
heck I know I do this with some american stuff
People don't seem to be regularly dying from peptides, at least not now.
If these substances actually had substantial effects on the body, people using them without medical supervision would almost certainly be overdosing with some frequency, I think.
(As I understand it, it is very possible to overdose on Ozempic… the reason it doesn't happen is because the purity is controlled, and people are taking specific medically-recommended doses.)
Don't take insulin if you can avoid it. Don't take other stupid shit if you can avoid it. You can definitely avoid this stupid shit.