While AI is incredible and has many usages, it is so confusing and produces so much "slop" (bad quality outputs) because of lack of proper education among the population that instead of me being without a job, people ask me if the AI is right on its takes.
In tech (or even biotech i this instance) GIGO still means "garbage in = garbage out". 😑
Just as the same issue exists in ALL technocratic science at the moment and looking ahead into the future- its a PEBCAP problem😐 = "Problem Exists Between Chair and PC", being the person inputting the data. $#!@ algorithms are still $#!@ whether done by AI or not, because AI is only using data acquired by the clever idiot that programmed the algorithms, even the LLMs. They cant "learn" on $#!@, and unfortunately or actually BECAUSE of our tech, WE have been getting stupider, not smarter.😐🤦♀️🤦♀️
GeneOptimizer (Raab sliding window algo) is more deterministic, avoids dataset biases, it more efficient when you use vast sequences, is more accurate in codon stability and immunogenicity, and balanced GC content to prevent misfolding (probably why the proline was added, and where), motif elimination (TLR-activating CpG sequences ones). Also more "scale friendly).
They would need a highly sophisticated epitope prediction tool and experimental validation to avoid the downsides, which you and I both know they skipped that part lol.
I'm thinking that, in spite of the medical freedom groups going on a freak out/betrayal-baying- at-the-moon rant, this AI is for mostly surveillance. What do you think? What advantage is there to have these huge AI facilities? How can it help us in day to day living? If it could stop child trafficking and do away with all the crooks, I'd be all in, but so far I am not aware of personal benefit.
Just in -- Trump admin supposedly working on executive order halt to federal funding of GOF.
Surveillance was overly present and abundant before AI, AI will just enable the analysis of all the data faster, and new ways to surveil of course, but mostly it is an architecture problem.
Current Machine Learning and AI architecture is AWFULLY INNEFICIENT. We are, quite literally, brute forcing our way out of problems, scalding up more and more because there are not enough algorithmic and architectural breakthroughs. Widespread use of AI, and especially nation-state level usage will require massive data centers under current conditions.
GoF will be exported to allied countries as it has been done for the last decades, it will avoid oversight, no stopping it in the age of AI too.
As venoms and toxins are being used in medicines (Gila Monster venom weight loss, snake venom blood thinners, cone snail venom opioid pain substitute, etc.), 🌽sider the similarities between the Ordeal Bean and the mRNA myocarditis clot shots. Calabar Ordeal Beans were given to prisoners to ascertain guilt or innocence in witch hunt drowning style. It the prisoners died, they were guilty. And the poison was a slow acting heart attack. Those who survived the bioweapons in Cov19 and the biosynth modified mRNA were “innocent” of weaker genes for TransHumanist Eugenics trying to Cull the Herd of weaker guilty genes. Dying straight off. Or a more slow acting myocarditis poisoned “died suddenly” death.
I've bookmarked this for future reference. Thank you, it's going to be super helpful to me. I'll look into getting your book when it's published! The work effort is astounding so I wish you great future success!
I keep reminding myself about Crichton's "life finds a way" and the antibiotics vs resistant bacteria debacle.
Suppose it is possible to create mRna medicines and tailor them down to individuals or small homogeneous groups, what are the risks and/or unintended consequences?
In the sci-fi worlds we find descriptions of species with faulty DNA, variations of breakdown, because of upstream manipulation.
Did you mean using mRNA as a "racial bioweapon" ? The risks would be off-target targeting such as someone sharing the specific "whatever" you are targeting, such as genes, enzymes, etc.
mRNA is to specific so it wouldn't "spill over", a virus, bacteria, a pathogen, well, exactly what you would expect, adaptation and getting out of hand and starting to target everyone of the same species.
I meant spillover and/or the target disease mutating into a new variant, because of the treatment protocol, which then becomes harder to treat or resistant to the mRNA treatment. So comparing with antibiotics, the over treatment and use has led to (virus? Bacteria?) strains resistant to the antibiotics (so then combo treatments). So by analogy, mRNA treatment creates a variant resistant to the treatment etc
I didn't think in terms of weaponizing, although I'd imagine it'd be a concern.
I just think everything has a cost and nature eventually sends the bill.
Oh that makes more sense. Cancer doesn't mutate or change as fast as pathogens, so using the immune system to target it is "safer" than, lets say, using mRNA to treat a respiratory infection, which the literature in 2020 and the creators were already saying "mRNA is not really a thing for treating infections, it is more suited for autoimmune diseases, and cancer". It was, from the ground up, a platform designed, specifically, to try and cure cancer.
Everything in it makes sense to treat cancer, rather than anything else. In this case, if the cancer survives, you take a sample, sequence it, and just target the new antigens (called neoantigens).
But, your last comment stands. Everything has a cost and nature eventually sends the bill. The question here is the cost (which I think it will/may be accelerate neurodegeneration in like 1 to 2 decades, PURELY A GUESS/gut instinct, based on nothing so far)
How will AI help them avoid the mistakes of last time in terms of LNP proliferation, cDNA creation, deposition of spike in endothelial cells, heart, testes, ovaries etc.?
They have no data at physiological level to model dynamics, so how will any models save them?
They had an opportunity where they could have allowed autopsies, which would have been equivalent of a time series as people dropped at different points in the journey of the spike in the bloodstream. But they blocked that.
So they don't have direct data.
The VAERS and the other app that they canned have indirect data, so they have to mine that first, set up hypotheses, and then simulate them in software. Even if they reproduce symptoms it does not mean they have the actual cause. How many years will this take even for the pharma industry?
On top of this, deep learning models need huge amounts of data.
Now, overlay the tech bros who have no idea what they are getting into. How many decades will that take?
Even if these bros are just pipes, why would they expect pharma to have not started their AI infrastructure already? From what you mention, they are ahead of these people.
Your first question. The deep learning mRNA distribution paper. They now can do it at a single cell level,
And I remember, clear as day both Pfizer and Moderna CEO proudly announcing they would be able to track each vaccine in each arm.
I have argued since 2022, they have excessive amounts of data and had no way to use it. Biotech has enough data to schedule all data centers in the US for dozens of years.
The project sounds desperate play from OpenAI hehe.
Moriarty, there is a different explanation that works. Assume the "story" about these companies creating the vaxxxines so quickly is what they decided to say to cover that they actually got the procedural cookbook from the bio-labs in Ukraine that were not only developing the C19 virus and its spike but also the vaxxxine design and production process. Given that C19 itself was Part 1 of a binary bioweapon-in-time, those wanting to spread that thing would want it to be the excuse to push a much more serious Part 2 bioweapon-in-time vaxxxine. They would have developed both as a tailored set. So your argument that the Pharma AIs chose the spike protein because of the RBD would be replaced by the idea that the bio-lab folks deliberately *chose* the spike protein because they wanted the maiming and the deaths. They then gave that idea and the necessary production process to Big Pharma to scale up (which they did poorly, but the GoF folks releasing C19 would have seen the contamination as a bonus, probably). In any case, I'm sure the Pharma companies involved did have AI software, but I am contending the original data came from the bio-labs along with the design of the vaxxxine to be made and the way to make it. Your approach assumes stupidity on the part of Pharma. My approach assumes malice on the part of those who designed and released C19 *and* stupidity and greed on the part of the Pharma companies. Too many things about C19 et al go in the wrong direction not to assume malice on *someone's* part. The press conference with Trump, Ellison, Son, and Altman was classic Trump-putting-bad-people-into-the-public-eye. Elon said later on X "They don't have the money." He despises Altman (perhaps rightfully). Ellison's hotel on Lake Tahoe just closed to events for some reason, and Son has been playing games for a long time and somehow staying afloat. Once someone online pointed out the body language, I started laughing.
The entire thing has a substantial, and extensive track record between the UNC > Antarctica (not substantial) > China, just as Western arrogance led to the current humiliation of an AI lab spending 5 million dollars and surpassing billions of dollars in Western AI, the same applies to, quite literally, every facet of industry and field of knowledge.
You can literally tweek the algorithm I share, and actually run it, you just need to acquire a few (public) datasets, it is not resource intensive, using efficient algos from a decade ago (that are still used today, proven track record) will get you similar results than those of Pfizer/Moderna/BioNtech.
Their algorithms chose the Spike Protein because using any antigenic prediction, or immunological algorithm for vaccine designs gets you the Spike, that thing will shine like 100 burning suns because of the superantigenic sites, the GP120 mimics, the RBD and its 15 mimics of other pathogens.
Pfizer and Moderna all moved fast, acquired extensive data analytics and data companies early on, and went on.
I don't assume stupidity, Westerns are just arrogant and think only "they can do things".
Having the time of my life with my favorite AI lab, DeepSeek basically destroying all the misconceptions in AI that I have been warning about for years.
Small Chinese AI developer that released, open source, free to use, what is right now THE MOST advanced AI, which in turn is advancing the entire field. It is faster, it is smarter, better, superior at everything. More creative, SIGNIFICANTLY LESS CENSORED M
They are hardware constrained, 2000 GPUs, and spent 5 million dollars to train this model. Compare to Open AI 50.000 of the best GPUs and hundreds of millions.
The devs are on Twitter and are super humble. Contrast to the sheer arrogance of Western AI devs.
I've been waiting for some small group to figure out how to make 50K-GPU server centers obsolete. Looks like you found something good. Is it DeepSeek.com for use by anyone?
While AI is incredible and has many usages, it is so confusing and produces so much "slop" (bad quality outputs) because of lack of proper education among the population that instead of me being without a job, people ask me if the AI is right on its takes.
Lol, lmao even, I got that one wrong.
In tech (or even biotech i this instance) GIGO still means "garbage in = garbage out". 😑
Just as the same issue exists in ALL technocratic science at the moment and looking ahead into the future- its a PEBCAP problem😐 = "Problem Exists Between Chair and PC", being the person inputting the data. $#!@ algorithms are still $#!@ whether done by AI or not, because AI is only using data acquired by the clever idiot that programmed the algorithms, even the LLMs. They cant "learn" on $#!@, and unfortunately or actually BECAUSE of our tech, WE have been getting stupider, not smarter.😐🤦♀️🤦♀️
#bluelightbrains #follownone #mistakeswereNOTmade #getlocalised
As for Pfizer using AI to design and codon-optimize BNT162b2, there were 2 versions admitted to (maybe there were versions 1-7?)
version 8 which used a BioNTech proprietary algorithm
version 9 which used a published algorithm (Raab et al. 2010)
they decided on version 9.....I have always wondered why
GeneOptimizer (Raab sliding window algo) is more deterministic, avoids dataset biases, it more efficient when you use vast sequences, is more accurate in codon stability and immunogenicity, and balanced GC content to prevent misfolding (probably why the proline was added, and where), motif elimination (TLR-activating CpG sequences ones). Also more "scale friendly).
They would need a highly sophisticated epitope prediction tool and experimental validation to avoid the downsides, which you and I both know they skipped that part lol.
Thanks for that explanation! I just thought it was ironic that a published algorithm worked better than their proprietary one. As for the rest…..yeah.
They probably ran a bunch of simulations or actual runs (burning R&D money and time, and compute !!!) and got better results from it.
It has a proven track record, it is reliable and very flexible.
I'm thinking that, in spite of the medical freedom groups going on a freak out/betrayal-baying- at-the-moon rant, this AI is for mostly surveillance. What do you think? What advantage is there to have these huge AI facilities? How can it help us in day to day living? If it could stop child trafficking and do away with all the crooks, I'd be all in, but so far I am not aware of personal benefit.
Just in -- Trump admin supposedly working on executive order halt to federal funding of GOF.
Surveillance was overly present and abundant before AI, AI will just enable the analysis of all the data faster, and new ways to surveil of course, but mostly it is an architecture problem.
Current Machine Learning and AI architecture is AWFULLY INNEFICIENT. We are, quite literally, brute forcing our way out of problems, scalding up more and more because there are not enough algorithmic and architectural breakthroughs. Widespread use of AI, and especially nation-state level usage will require massive data centers under current conditions.
GoF will be exported to allied countries as it has been done for the last decades, it will avoid oversight, no stopping it in the age of AI too.
As venoms and toxins are being used in medicines (Gila Monster venom weight loss, snake venom blood thinners, cone snail venom opioid pain substitute, etc.), 🌽sider the similarities between the Ordeal Bean and the mRNA myocarditis clot shots. Calabar Ordeal Beans were given to prisoners to ascertain guilt or innocence in witch hunt drowning style. It the prisoners died, they were guilty. And the poison was a slow acting heart attack. Those who survived the bioweapons in Cov19 and the biosynth modified mRNA were “innocent” of weaker genes for TransHumanist Eugenics trying to Cull the Herd of weaker guilty genes. Dying straight off. Or a more slow acting myocarditis poisoned “died suddenly” death.
https://naturespoisons.com/2014/04/15/physostigmine-from-ordeal-poison-to-valuable-medicine-calabar-bean/
I've bookmarked this for future reference. Thank you, it's going to be super helpful to me. I'll look into getting your book when it's published! The work effort is astounding so I wish you great future success!
👍👍👍
I keep reminding myself about Crichton's "life finds a way" and the antibiotics vs resistant bacteria debacle.
Suppose it is possible to create mRna medicines and tailor them down to individuals or small homogeneous groups, what are the risks and/or unintended consequences?
In the sci-fi worlds we find descriptions of species with faulty DNA, variations of breakdown, because of upstream manipulation.
Did you mean using mRNA as a "racial bioweapon" ? The risks would be off-target targeting such as someone sharing the specific "whatever" you are targeting, such as genes, enzymes, etc.
mRNA is to specific so it wouldn't "spill over", a virus, bacteria, a pathogen, well, exactly what you would expect, adaptation and getting out of hand and starting to target everyone of the same species.
I meant spillover and/or the target disease mutating into a new variant, because of the treatment protocol, which then becomes harder to treat or resistant to the mRNA treatment. So comparing with antibiotics, the over treatment and use has led to (virus? Bacteria?) strains resistant to the antibiotics (so then combo treatments). So by analogy, mRNA treatment creates a variant resistant to the treatment etc
I didn't think in terms of weaponizing, although I'd imagine it'd be a concern.
I just think everything has a cost and nature eventually sends the bill.
Oh that makes more sense. Cancer doesn't mutate or change as fast as pathogens, so using the immune system to target it is "safer" than, lets say, using mRNA to treat a respiratory infection, which the literature in 2020 and the creators were already saying "mRNA is not really a thing for treating infections, it is more suited for autoimmune diseases, and cancer". It was, from the ground up, a platform designed, specifically, to try and cure cancer.
Everything in it makes sense to treat cancer, rather than anything else. In this case, if the cancer survives, you take a sample, sequence it, and just target the new antigens (called neoantigens).
But, your last comment stands. Everything has a cost and nature eventually sends the bill. The question here is the cost (which I think it will/may be accelerate neurodegeneration in like 1 to 2 decades, PURELY A GUESS/gut instinct, based on nothing so far)
'personalized medicine' in a form of SYNTHETIC GENETIC materials is called GENE THERAPY. It overwrites your ORIGINAL HUMAN GENOME.
Why don't you finally change your language in order to really help UNEDUCATED people to completely reject that lethal, lifeless technology???
How will AI help them avoid the mistakes of last time in terms of LNP proliferation, cDNA creation, deposition of spike in endothelial cells, heart, testes, ovaries etc.?
They have no data at physiological level to model dynamics, so how will any models save them?
They had an opportunity where they could have allowed autopsies, which would have been equivalent of a time series as people dropped at different points in the journey of the spike in the bloodstream. But they blocked that.
So they don't have direct data.
The VAERS and the other app that they canned have indirect data, so they have to mine that first, set up hypotheses, and then simulate them in software. Even if they reproduce symptoms it does not mean they have the actual cause. How many years will this take even for the pharma industry?
On top of this, deep learning models need huge amounts of data.
Now, overlay the tech bros who have no idea what they are getting into. How many decades will that take?
Even if these bros are just pipes, why would they expect pharma to have not started their AI infrastructure already? From what you mention, they are ahead of these people.
This entire project sounds not well thoughtout.
Your first question. The deep learning mRNA distribution paper. They now can do it at a single cell level,
And I remember, clear as day both Pfizer and Moderna CEO proudly announcing they would be able to track each vaccine in each arm.
I have argued since 2022, they have excessive amounts of data and had no way to use it. Biotech has enough data to schedule all data centers in the US for dozens of years.
The project sounds desperate play from OpenAI hehe.
Moriarty, there is a different explanation that works. Assume the "story" about these companies creating the vaxxxines so quickly is what they decided to say to cover that they actually got the procedural cookbook from the bio-labs in Ukraine that were not only developing the C19 virus and its spike but also the vaxxxine design and production process. Given that C19 itself was Part 1 of a binary bioweapon-in-time, those wanting to spread that thing would want it to be the excuse to push a much more serious Part 2 bioweapon-in-time vaxxxine. They would have developed both as a tailored set. So your argument that the Pharma AIs chose the spike protein because of the RBD would be replaced by the idea that the bio-lab folks deliberately *chose* the spike protein because they wanted the maiming and the deaths. They then gave that idea and the necessary production process to Big Pharma to scale up (which they did poorly, but the GoF folks releasing C19 would have seen the contamination as a bonus, probably). In any case, I'm sure the Pharma companies involved did have AI software, but I am contending the original data came from the bio-labs along with the design of the vaxxxine to be made and the way to make it. Your approach assumes stupidity on the part of Pharma. My approach assumes malice on the part of those who designed and released C19 *and* stupidity and greed on the part of the Pharma companies. Too many things about C19 et al go in the wrong direction not to assume malice on *someone's* part. The press conference with Trump, Ellison, Son, and Altman was classic Trump-putting-bad-people-into-the-public-eye. Elon said later on X "They don't have the money." He despises Altman (perhaps rightfully). Ellison's hotel on Lake Tahoe just closed to events for some reason, and Son has been playing games for a long time and somehow staying afloat. Once someone online pointed out the body language, I started laughing.
The entire thing has a substantial, and extensive track record between the UNC > Antarctica (not substantial) > China, just as Western arrogance led to the current humiliation of an AI lab spending 5 million dollars and surpassing billions of dollars in Western AI, the same applies to, quite literally, every facet of industry and field of knowledge.
You can literally tweek the algorithm I share, and actually run it, you just need to acquire a few (public) datasets, it is not resource intensive, using efficient algos from a decade ago (that are still used today, proven track record) will get you similar results than those of Pfizer/Moderna/BioNtech.
Their algorithms chose the Spike Protein because using any antigenic prediction, or immunological algorithm for vaccine designs gets you the Spike, that thing will shine like 100 burning suns because of the superantigenic sites, the GP120 mimics, the RBD and its 15 mimics of other pathogens.
Pfizer and Moderna all moved fast, acquired extensive data analytics and data companies early on, and went on.
I don't assume stupidity, Westerns are just arrogant and think only "they can do things".
Having the time of my life with my favorite AI lab, DeepSeek basically destroying all the misconceptions in AI that I have been warning about for years.
OK. What is DeepSeek?
Small Chinese AI developer that released, open source, free to use, what is right now THE MOST advanced AI, which in turn is advancing the entire field. It is faster, it is smarter, better, superior at everything. More creative, SIGNIFICANTLY LESS CENSORED M
They are hardware constrained, 2000 GPUs, and spent 5 million dollars to train this model. Compare to Open AI 50.000 of the best GPUs and hundreds of millions.
The devs are on Twitter and are super humble. Contrast to the sheer arrogance of Western AI devs.
I've been waiting for some small group to figure out how to make 50K-GPU server centers obsolete. Looks like you found something good. Is it DeepSeek.com for use by anyone?
chat.deepseek.com
Yes free to everyone
Cool. Thanks!