Bioenergetic AI Labs: The future of Ray Peat's legacy is already set
-
@sharko @yerrag @Peatful @Kvirion
I would like to use this tech if it returned links/citations back to source material. I think it's a cool experiment and fine to ask to be paid for the archival value-add that the LLM (might) provide.I guess, @sharko, you are not a quant stats modeler who understands the "stochastic parrot" comment by @Kvirion and paper he linked to or valid concerns raised by @Peatful
The prompts you posted telling your LLM to "invest 100 times more" and do something "even more surprising" to Peat, etc., reveals, to me that you're unaware how those prompts are parsed and what's happening under the hood (in the code). The code will assemble a ranked list of word associations and return the top-ranked test composition by some very limited likelihood criterion in its very "small-world" model space. When you tell it to "come up with an idea" (it cannot!...it can only assemble combinations it was trained on!)...it simply returns #2 on the ranked list or does a re-ranking with a final check to give you something else on its list of guesses/synthetic text assemblages.
The idea that "AI" is "getting smarter" or graduating from high-school- to PhD-level "thinking" is absurd. This tech can be a tremendous aid for constructing text or computer code or many other useful applications, where the training dataset is sufficient in some sense to satisfy the task at hand. Thus, as an archive tool that can spot likely interconnections in the literature or sources you trained it on, I agree there is a real value-add.
When it comes to synthesizing new ideas as in your last few posts, any claim of "synthesizing new ideas" falls flat and does become (borderline?) dishonest. It would be dishonest, for example, to claim that what is returned by your LLM represents a "likely" Ray Peat take on a new topic he never wrote about.
Keep in mind that the training dataset is a binding constraint. The archival value-add is genuine. Claiming that the LLM comes up with "new insights" in any way associated with Ray Peat is dishonest in my view. Better to say explicitly "new insights generated by an LLM trained on a specialized training set comprised of..." and the list precisely which sources were included in the training set. Ray Peat's, Broda Barnes' oeuvres. Some filtered pub-med content you've evidently included. That specificity would match the archival value-add claim. I recommend strongly against selling the LLM with the repeated prompts you gave upthread, which, to anyone working in this space is unlikely to impress.
-
@T-3 said in Bioenergetic AI Labs: The future of Ray Peat's legacy is already set:
@sharko @yerrag @Peatful @Kvirion
I would like to use this tech if it returned links/citations back to source material. I think it's a cool experiment and fine to ask to be paid for the archival value-add that the LLM (might) provide.I guess, @sharko, you are not a quant stats modeler who understands the "stochastic parrot" comment by @Kvirion and paper he linked to or valid concerns raised by @Peatful
The prompts you posted telling your LLM to "invest 100 times more" and do something "even more surprising" to Peat, etc., reveals, to me that you're unaware how those prompts are parsed and what's happening under the hood (in the code). The code will assemble a ranked list of word associations and return the top-ranked test composition by some very limited likelihood criterion in its very "small-world" model space. When you tell it to "come up with an idea" (it cannot!...it can only assemble combinations it was trained on!)...it simply returns #2 on the ranked list or does a re-ranking with a final check to give you something else on its list of guesses/synthetic text assemblages.
The idea that "AI" is "getting smarter" or graduating from high-school- to PhD-level "thinking" is absurd. This tech can be a tremendous aid for constructing text or computer code or many other useful applications, where the training dataset is sufficient in some sense to satisfy the task at hand. Thus, as an archive tool that can spot likely interconnections in the literature or sources you trained it on, I agree there is a real value-add.
When it comes to synthesizing new ideas as in your last few posts, any claim of "synthesizing new ideas" falls flat and does become (borderline?) dishonest. It would be dishonest, for example, to claim that what is returned by your LLM represents a "likely" Ray Peat take on a new topic he never wrote about.
Keep in mind that the training dataset is a binding constraint. The archival value-add is genuine. Claiming that the LLM comes up with "new insights" in any way associated with Ray Peat is dishonest in my view. Better to say explicitly "new insights generated by an LLM trained on a specialized training set comprised of..." and the list precisely which sources were included in the training set. Ray Peat's, Broda Barnes' oeuvres. Some filtered pub-med content you've evidently included. That specificity would match the archival value-add claim. I recommend strongly against selling the LLM with the repeated prompts you gave upthread, which, to anyone working in this space is unlikely to impress.
I think you say mostly correct things but:
First, the correspondences I sent are an interesting game. I've been playing with AI since the day Openai launched ChatGPT and from the first days I "broke" almost every limitation openai tried to put on it and made it do what I wanted.
It is true that the banal and excessive requests I asked of him will not really allow him to "invest 100 times more", but he does have a (very average) ability to understand that if I ask him for something "crazy", he should find something that can be considered extraordinary.
On the other hand, he is also stupid to the same extent.
The essence of these tools is really to speed up research processes and I don't use them for the examples I sent, it's just "cool" in my opinion.
Did you go through his answers? I didn't pretend to say he was coming up with new ideas that Ray Peat hadn't thought of, I just pasted the prompt so they knew what I asked him.
This particular tool works with embeddings and chunks of 3,000 words. He looks for 5 chunks most relevant to the phrases he chooses from the prompt and runs one of the models I choose (gpt4o in this case) on them. Which means, he chooses 15,000 relevant words and uses the rest of the vast knowledge he's trained on, along with access to a Google search, and compiles the answer.
It is not true to say that he cannot invent new things. According to the process I mentioned, he can manage to assemble several edges to "treat" a certain issue. Most of the knowledge that Ray spread, does not go into detail about treatments, if you for example read all his content and listened to all the interviews, you will draw the conclusions you are able to draw from it, the AI is also able to "put together" conclusions (very quickly)...
Regarding your question about sources: right now I can say that it contains maybe 95% of all the information that Ray distributed. He cannot attach sources to each answer, because chunks will not necessarily have clear information about the source. But you can ask him, for example, for an exact quote relevant to each section. Sometimes he will also be able to provide the name of an interview, newsletter, article or book.
Although the examples I sent are really not the main purpose of using it, it's very fun to play with it and the better you get to know Ray's works, you can get to very interesting places with it this way (through a more professional conversation of course)
-
Me:
Create a picture of the pathways of T3 in the body
Him:
Here is the diagram illustrating the pathways of T3 in the body, including the intracellular pathways such as its interaction with mitochondria. It shows the thyroid gland producing T4, its conversion to T3, and the movement and effects of T3 throughout the body, highlighting its role in energy production and metabolism.
Of course the picture is not accurate. Soon these technologies will know how to produce accurate images as well as videos. This will allow us to understand the studied material on a completely different level
-
I've been looking at this nightmarish "diagram" for far too long.
-
@NNight said in Bioenergetic AI Labs: The future of Ray Peat's legacy is already set:
I've been looking at this nightmarish "diagram" for far too long.
It is not part of the system because as I noted, public AI technologies are still not yet able to produce images of this kind at a satisfactory level. This was an example of another extremely useful future AI capability, for all the AI naysayers out there
-
@sharko
I'm not a naysayer, more like a realist about capabilities of AI.
I agree with T-3, you don't seem to really understand how these models work (your prompt are a bit "fanciful"). And I'm not saying that to be mean.Ps: I think LLM can be useful for translation, summarization, categorization.
-
Some shitty AI parroting his words taken out of context is a joke and not his legacy. So many people trying to profit off of his name since he died. Maybe at least try and sound less like a pompous ass by not calling this his legacy? AI incapable of own thought, reason and creativity isn't his legacy.
-
@NNight said in Bioenergetic AI Labs: The future of Ray Peat's legacy is already set:
@sharko
I'm not a naysayer, more like a realist about capabilities of AI.
I agree with T-3, you don't seem to really understand how these models work (your prompt are a bit "fanciful"). And I'm not saying that to be mean.Ps: I think LLM can be useful for translation, summarization, categorization.
This is a fairly common attitude among Ray fans, many of us like to skip information in order to continue to stick to our faith. You ignored the response I wrote to him. Anyway, I'm really not interested in trying to convince anyone. I enjoy technology that helps me help myself and others.
-
@Barghest said in Bioenergetic AI Labs: The future of Ray Peat's legacy is already set:
Some shitty AI parroting his words taken out of context is a joke and not his legacy. So many people trying to profit off of his name since he died. Maybe at least try and sound less like a pompous ass by not calling this his legacy? AI incapable of own thought, reason and creativity isn't his legacy.
It's a common symptom of Ray fans who are sure they have all the answers to everything. You don't know me and you also happen to be completely wrong. If my main goal was money, first of all I wouldn't focus on such a small niche and if I did, I would make a bot that doesn't cost me money in half an hour and sell to all 20 thousand people who search for Ray Peat Diet every month. Hi, I gave advice to anyone who is trying to get rich from Ray Peat. I'll tell you more than that: your attitude is so negative that I doubt your organism is approaching homeostasis according to Ray Peat's approach. There is no problem in making money, if in the end it contributes to the world. Any other story stems from unclear thinking and a general lack of understanding
-
@sharko You really don't get what @T-3, @Peatful, me, and others are trying to explain to you, do you?
Maybe because you don't know what you don't know i.e. you're probably affected by Dunning–Kruger effect...
Your convictions/claims are mostly false, because of
- Epistemology:
LLMs are only capable of partial deductive and inductive reasoning in statistical ways.
They are unable to perform Creative (innovative/inventive) thinking i.e. abductive reasoning (e.g. "logic of hunches") - this is a scientific fact. - Ontology:
LLM's algorithms are linear and statistical, limited by datasets i.e. closed IT systems. But the human mind is a nonlinear, interrelated, adaptive, partially quantum open complex ecosystem. This is a big difference. - Phenomenology:
A human mind operates with a sense of self-awareness and intentionality, we perceive, think, and act, etc.
LLMs lack consciousness and intentionality. They generate responses based on patterns in the data they were trained on, without any subjective experience or self-awareness. Their outputs have no underlying intention or purpose; they just produce statistically probable text.
Moreover, human cognition is deeply embodied. Our thoughts and experiences are influenced by our physical bodies and sensory inputs. Emotions, physical sensations, and the environment play critical roles in shaping our mental states.
We are emotional and adaptable - continuous/dynamic improvement (or regress); can also reflect, and see things from different angles (at least some of us). LLMs do not. - Axiology: LLM doesn't get ethics...
Plus basics of Knowledge management: people can think more than they can say, and they can say more, they can write...
So, "AI" can help us find some useful info, but can not bring new ideas.
- Epistemology:
-
Thanks for sharing these ideas. Did you learn these concepts by self-study? Or did these come from a liberal arts education that are similar to Peat's that give such a perspective?
Can you be more constructive rather than talk down then as if you want to give a lecture more than help Sharko make his efforts of using AI to help us gain a better understanding of Ray Peat's work?
But I may be wrong in not giving you and T-3 enough credit as really I can see some effort to be constructive, and that Sharko's responses to you may not hit the right notes with you, though I get the sense that he is more about explaining the possibilities and potentials if AI, than in addressing the points you raised.
Sharko, I can see if I were in your shoes I would feel like I'm defending a thesis from a panel consisting of judges ranging from caustic and belligerent to constructive and helpful. It would make you uncomfortable and defensive if only for the belligerent trying to have his say in rhe tone he only is used to.
But let's continue the diacussion giving Sharko a chance to answer Kvirion's points, and if Kvirion's points are not sufficiently addressed in the current prototypical stage, then we have to consider the likelihood of improving the AI model instead of prejudice it based on previous attempts of AI that failed.
-
@Barghest said in Bioenergetic AI Labs: The future of Ray Peat's legacy is already set:
Some shitty AI parroting his words taken out of context is a joke and not his legacy. So many people trying to profit off of his name since he died. Maybe at least try and sound less like a pompous ass by not calling this his legacy? AI incapable of own thought, reason and creativity isn't his legacy.
I appreciate the objectivity / understanding here balanced with passion
Said much better than I have
Ray was a gifted soul of a man
In the legacy thread from the old forum iirc-
someone else, again, summarized it as digital vs analogAnd it really comes down to relationship
Relationship to yourself, others and the environment
Expressed intellectually and or creativelyPeat was the epitome of both
Soulless LLM or AI offers me nothing
-
@yerrag said in Bioenergetic AI Labs: The future of Ray Peat's legacy is already set:
Thanks for sharing these ideas. Did you learn these concepts by self-study? Or did these come from a liberal arts education that are similar to Peat's that give such a perspective?
A mix of... partial education in liberal arts plus self-study thanks to some wise people sharing their insights online/books... Plus recently reading/learning from Ray's works (I miss him so much...) helped me to add another dimension... but I'm still learning and I'm open to dialogue
BTW I also received professional training in advanced sense-making and Complex Adaptive Systems.
Can you be more constructive rather than talk down then as if you want to give a lecture more than help Sharko make his efforts of using AI to help us gain a better understanding of Ray Peat's work?
Yeah, you're right, my bad. I may try... but both sides need to be more open-minded...
Honestly, I'm pretty frustrated with the IT guys claiming unfoundedly that they have a panacea or they are helping the world... When in reality they are unaware (WEF/neoliberal?) agents of destruction/idiocracy (i.e. Moloch)...In my imagination I'm with John Connor, Morpheus, and Butlerian Jihad - fighting the machines!
BTW I tried to be nice to Sharko at first...
Full disclosure - in the ancient past I also worked in roles of IT database developer/Analyst or IT project manager and I was a technology fanboy.
But I may be wrong in not giving you and T-3 enough credit as really I can see some effort to be constructive, and that Sharko's responses to you may not hit the right notes with you, though I get the sense that he is more about explaining the possibilities and potentials if AI, than in addressing the points you raised.
Right, it's good to use/explore possibilities and potentials, BUT one also must be aware of (many) limitations...
LLM can help us find something, but such info must not be seen as a conclusion, but only as an input for further conscious processing with the help of the scientific method and creativity...
But let's continue the diacussion giving Sharko a chance to answer Kvirion's points, and if Kvirion's points are not sufficiently addressed in the current prototypical stage, then we have to consider the likelihood of improving the AI model instead of prejudice it based on previous attempts of AI thst failed.
Golden advice, I'm for it.
-
@sharko said in Bioenergetic AI Labs: The future of Ray Peat's legacy is already set:
@NNight said in Bioenergetic AI Labs: The future of Ray Peat's legacy is already set:
@sharko
I'm not a naysayer, more like a realist about capabilities of AI.
I agree with T-3, you don't seem to really understand how these models work (your prompt are a bit "fanciful"). And I'm not saying that to be mean.Ps: I think LLM can be useful for translation, summarization, categorization.
This is a fairly common attitude among Ray fans, many of us like to skip information in order to continue to stick to our faith. You ignored the response I wrote to him. Anyway, I'm really not interested in trying to convince anyone. I enjoy technology that helps me help myself and others.
I'm not really a "Ray fan" and I've always despised the cultish behavior that I had observed on the RPF.
My goal was to give you an honest feedback on your business model (and to criticize some of your "idealism").
I don't want to appear as someone who want to discourage you, this is not my goal. I think it's very good that you want to develop such a project and I'm not of those who are afraid of AI. Again, good Luck!
-
@Kvirion said in Bioenergetic AI Labs: The future of Ray Peat's legacy is already set:
@sharko You really don't get what @T-3, @Peatful, me, and others are trying to explain to you, do you?
Maybe because you don't know what you don't know i.e. you're probably affected by Dunning–Kruger effect...
Your convictions/claims are mostly false, because of
- Epistemology:
LLMs are only capable of partial deductive and inductive reasoning in statistical ways.
They are unable to perform Creative (innovative/inventive) thinking i.e. abductive reasoning (e.g. "logic of hunches") - this is a scientific fact. - Ontology:
LLM's algorithms are linear and statistical, limited by datasets i.e. closed IT systems. But the human mind is a nonlinear, interrelated, adaptive, partially quantum open complex ecosystem. This is a big difference. - Phenomenology:
A human mind operates with a sense of self-awareness and intentionality, we perceive, think, and act, etc.
LLMs lack consciousness and intentionality. They generate responses based on patterns in the data they were trained on, without any subjective experience or self-awareness. Their outputs have no underlying intention or purpose; they just produce statistically probable text.
Moreover, human cognition is deeply embodied. Our thoughts and experiences are influenced by our physical bodies and sensory inputs. Emotions, physical sensations, and the environment play critical roles in shaping our mental states.
We are emotional and adaptable - continuous/dynamic improvement (or regress); can also reflect, and see things from different angles (at least some of us). LLMs do not. - Axiology: LLM doesn't get ethics...
Plus basics of Knowledge management: people can think more than they can say, and they can say more, they can write...
So, "AI" can help us find some useful info, but can not bring new ideas.
You're writing things down that sound like I once said that my goal is for AI to replace us.
I have explained several times that these are tools that can accelerate research processes at a dizzying pace and this is the essence of what I am offering.
I'm not an AI freak, I was a temp when it came out and ran away from it once I couldn't control my serotonin even with substances. (This is what happens when you try to solve problems in your head intensively without a break).
I will ask you a question:
Suppose you are researching according to the knowledge of Ray Peat but you do not have a basic understanding of human physiology and biology, how long would it take you to understand the picture if you were to read all of Ray Peat's information and at the same time learn basic things about human physiology and biology through Google, Pubmed, etc. and how long would it take you if you used an AI that contained all the knowledge that Ray Peat published, along with a lot of additional knowledge and unlimited access to research?
Another question:
If you were to ask an AI agent that is capable of performing several processes one after the other in response to a prompt, connected to a Vector DB containing chunks of 3,000 words and structured so that it is possible to quickly retrieve the 5 chunks most relevant to the user's prompt, after being filtered and improved by gpt-4o and returning An answer after analyzing 15,000 words with the most relevant content of all the knowledge Ray Peat published in a clear language of your choice, checking it against relevant external information and information he was trained on by gpt-4o, with let's say Gemini 1.5 flash model (because we want a high token model and 1M token input+output is better than the others for the part it need to check all the collected data together)
“The goal: to find the most likely reasons why some thyroid supplements work and others don't.
Go over your knowledge and extract the most relevant information about:
T4 (Thyroxine) T3 (Triiodothyronine) rT3 (Reverse Triiodothyronine) T2 (Diiodothyronine) T1 (Monioiodothyronine) Calcitonin
Then, you will work on the knowledge you have been trained on and extract relevant information about each of these hormones.
After that, you will search pubmed for 5-10 studies that contain information about each of these hormones, for each of them and summarize the 5-10 most relevant findings for each of the hormones.
Then, check 3 studies on each of the following list of drugs and supplements: *** and summarize the findings indicating their effect, along with information on their ingredients if any.
After that, go through all the information you have gathered, provide a list of possible and accurate reasons why a particular drug or supplement affects body temperature and heart rate and others do not, relevant quotes from which you drew the conclusions and links to the sources of information for the relevant quotes."
Do you think there is no way he will find new insights that no one had found before?
- Epistemology:
-
@NNight said in Bioenergetic AI Labs: The future of Ray Peat's legacy is already set:
@sharko said in Bioenergetic AI Labs: The future of Ray Peat's legacy is already set:
@NNight said in Bioenergetic AI Labs: The future of Ray Peat's legacy is already set:
@sharko
I'm not a naysayer, more like a realist about capabilities of AI.
I agree with T-3, you don't seem to really understand how these models work (your prompt are a bit "fanciful"). And I'm not saying that to be mean.Ps: I think LLM can be useful for translation, summarization, categorization.
This is a fairly common attitude among Ray fans, many of us like to skip information in order to continue to stick to our faith. You ignored the response I wrote to him. Anyway, I'm really not interested in trying to convince anyone. I enjoy technology that helps me help myself and others.
I'm not really a "Ray fan" and I've always despised the cultish behavior that I had observed on the RPF.
My goal was to give you an honest feedback on your business model (and to criticize some of your "idealism").
I don't want to appear as someone who want to discourage you, this is not my goal. I think it's very good that you want to develop such a project and I'm not of those who are afraid of AI. Again, good Luck!
Thank you.
It's really just a side project and my main goal is to recruit friends for the research tools, so that I can continue to use them myself and develop them, for the sake of my study goals and at the same time, if it ends up generating profits for me, it's only will allow me to focus more on that, at the expense of other things I do today, like selling digital health courses that make me enough money so that I don't have to spend time on bots for extra income.
It's so unprofitable compared to all my alternatives, that it's a little funny to me that there are those who think I'm trying to ride on someone's knowledge to make money. And what's more: that those people think that riding on someone's knowledge to make money by using AI technologies is bad, but learning from all of their knowledge and making a living helping others with their health, thanks to this knowledge - is fine.
-
@sharko said in Bioenergetic AI Labs: The future of Ray Peat's legacy is already set:
I will ask you a question:
Suppose you are researching according to the knowledge of Ray Peat but you do not have a basic understanding of human physiology and biology, how long would it take you to understand the picture if you were to read all of Ray Peat's information and at the same time learn basic things about human physiology and biology through Google, Pubmed, etc. and how long would it take you if you used an AI that contained all the knowledge that Ray Peat published, along with a lot of additional knowledge and unlimited access to research?
Sorry, but with all the respect, you are making the same mistake over and over...
The theory of cognitive Predictive Processing claims that "People don't see, what they do not expect to see". You seem to be a great example of it...
Wisdom isn't about (the speed of) information processing.
It is about the process of continuous sense-making... (learning, exploration, probing, understanding/framing, analyzing chunks, synthesizing contradictions, contextualizing, refining, questioning assumptions, epistemic humility, etc.)If one does not have a basic understanding of human physiology and biology (not to mention ontology, epistemology, phenomenology, and axiology) then one shouldn't even approach this field, or one may get seriously hurt...
Moreover, you seem to assume that all knowledge sources are equal and rational. And that a person is asking the right question...
Knowledge is like a forest - each component (trees, fungi/Mycorrhiza, insects, soil, rain, sun, prey/predators, season) and their relationships are important! But you seem to be only interested in counting kilograms of wood...
One should know [before] that a body is an ecosystem of interrelated components with synergistic effects, feedforward/feedback loops, allostatic/homeostatic mechanisms, etc. That most substances/substrates and their effects are context- and path-dependent. Also, understand what is the current problem with peer-reviewed publications. How to differentiate between a paper written with a reductionistic approach or a systemic (holistic one). What is the problem with p-values... And this is only level one...
Have you ever heard about Popper, Kuhn, and Lakatos, and how they define a theory, a paradigm (change), and the process of gaining understanding in general?
And all the points above are just the tip of the iceberg of scientific Knowledge Management...
You want to offer definitive answers in a non-definitive universe... This will not end well...
What you propose is a reductionistic point-based (not systemic) information stripped from all dependencies - we already have it in the mainstream and this is a key problem! What you propose is the opposite of what Ray was for...
If you were to ask an AI agent that is capable of performing several processes one after the other in response to a prompt, connected to a Vector DB containing chunks of 3,000 words and structured so that it is possible to quickly retrieve the 5 chunks most relevant to the user's prompt, after being filtered and improved by gpt-4o and returning An answer after analyzing 15,000 words with the most relevant content of all the knowledge Ray Peat published in a clear language of your choice, checking it against relevant external information and information he was trained on by gpt-4o
It's just soulless/blind information mining...
Where is Perceive-Think-Act?
-
@Kvirion said in Bioenergetic AI Labs: The future of Ray Peat's legacy is already set:
@sharko said in Bioenergetic AI Labs: The future of Ray Peat's legacy is already set:
I will ask you a question:
Suppose you are researching according to the knowledge of Ray Peat but you do not have a basic understanding of human physiology and biology, how long would it take you to understand the picture if you were to read all of Ray Peat's information and at the same time learn basic things about human physiology and biology through Google, Pubmed, etc. and how long would it take you if you used an AI that contained all the knowledge that Ray Peat published, along with a lot of additional knowledge and unlimited access to research?
Sorry, but with all the respect, you are making the same mistake over and over...
The theory of cognitive Predictive Processing claims that "People don't see, what they do not expect to see". You seem to be a great example of it...
Wisdom isn't about (the speed of) information processing.
It is about the process of continuous sense-making... (learning, exploration, probing, understanding/framing, analyzing chunks, synthesizing contradictions, contextualizing, refining, questioning assumptions, epistemic humility, etc.)If one does not have a basic understanding of human physiology and biology (not to mention ontology, epistemology, phenomenology, and axiology) then one shouldn't even approach this field, or one may get seriously hurt...
Knowledge is like a forest - each component (trees, fungi/Mycorrhiza, insects, soil, rain, sun, prey/predators, season) and their relationships are important! But you seem to be only interested in counting kilograms of wood...
One should know [before] that a body is an ecosystem of interrelated components with synergistic effects, feedforward/feedback loops, allostatic/homeostatic mechanisms, etc. That most substances/substrates and their effects are context- and path-dependent. Also, understand what is the current problem with peer-reviewed publications. How to differentiate between a paper written with a reductionistic approach or a systemic (holistic one). What is the problem with p-values... And this is only level one...
Moreover, you seem to assume that all knowledge sources are equal and rational. And that a person is asking the right question...
Have you ever heard about Popper, Kuhn, and Lakatos, and how they define a theory, a paradigm (change), and the process of gaining understanding in general?And all the above points are just the tip of the iceberg of Knowledge Management...
You want to offer definitive answers in a non-definitive universe... This will not end well
What you propose is a reductionistic point-based (not systemic) information stripped from all dependencies - we already have it in the mainstream and this is a key problem! What you propose is the opposite of what Ray was for...
If you were to ask an AI agent that is capable of performing several processes one after the other in response to a prompt, connected to a Vector DB containing chunks of 3,000 words and structured so that it is possible to quickly retrieve the 5 chunks most relevant to the user's prompt, after being filtered and improved by gpt-4o and returning An answer after analyzing 15,000 words with the most relevant content of all the knowledge Ray Peat published in a clear language of your choice, checking it against relevant external information and information he was trained on by gpt-4o
It's just soulless/blind information mining...
Your attitude is excellent but you tend to think that this is the only way...
I will tell you that I started studying Ray Peat's knowledge 10 years ago and to this day I lack about 95% of the knowledge of basic physiology and biology and despite this - in the last decade, I have brought myself back from the grave several times (the number of serious problems I had at the same time is greater than you can imagine ). A few examples out of maybe 100: for a year I would wake up every day after exactly 5 hours with massive chest pains and crazy sweating, along with hand tremors for 15 years, tingling all over my body and head, tics, neurological problems, manic depression, vision problems, severe stomach pains , 0 energy, crops and that's just the tip of the iceberg.
Every time after I pushed myself to the edge, I returned to a "new" state in less than three months, each time in a different and complex way, based on connecting a lot of dots according to the bioenergetic approach, without knowing almost anything about basic physiology and biology.
I don't ask AI a question, get an answer and say thank you. I check everything 100 times, fix it and improve its understanding as I go. In the end, I'm still learning at a speed 100 times more than it would have taken me through Google (as someone who has been an expert in Google search for over 25 years and of which 15 years as an seo expert who knows exactly what to choose from all the top false results and how to find the more correct answers).
I enjoy the discussion, you have no idea how insignificant it is for me to recruit customers for this bot compared to other things I do.
What's more, I understand if some of you were upset because of the title of the post and examples of correspondence with the AI that I sent for fun and made you raise an eyebrow.
Regarding the title: I'm sorry, the marketing comes out of me naturally without even noticing. habits...
-
@sharko said in Bioenergetic AI Labs: The future of Ray Peat's legacy is already set:
I don't ask AI a question, get an answer and say thank you. I check everything 100 times, fix it and improve its understanding as I go. In the end, I'm still learning at a speed 100 times more than it would have taken me through Google (as someone who has been an expert in Google search for over 25 years and of which 15 years as an seo expert who knows exactly what to choose from all the top false results and how to find the more correct answers).
Will your "AI" do that? Will ask a user to check different sources, think about different approaches, ask a user to experiment, ask a user to think of the context of an unhealthy condition and what preceded it?
And a key question: what is the chance that your ego is bigger than your knowledge?
-
@sharko said in Bioenergetic AI Labs: The future of Ray Peat's legacy is already set:
A few examples out of maybe 100: for a year I would wake up every day after exactly 5 hours with massive chest pains and crazy sweating, along with hand tremors for 15 years, tingling all over my body and head, tics, neurological problems, manic depression, vision problems, severe stomach pains , 0 energy, crops and that's just the tip of the iceberg.
It doesn't prove anything, some issues may be easier to fix, some are more difficult, and sometimes it's just luck... or confabulation...
Will you take responsibility, that someone asking your "favorite toy" for advice may die because your "AI" miscalculated a statistic between words...?