Thank you Alter.Systems team! Keep up the good work. Alter Ai has been instrumental in helping me manage my own medical care, and my family and pets as well. It provides the kind of methodical, wholistic assessment that has been completey removed from the US health care industry. Unless you can find a good concierge doc, which I haven't been able to find.
Steve, what is your personal connection to AlterAI and what do you hope to accomplish in the world with it, or see it accomplish once these hires are made and it's better equipped with more resources? And. . . why is this company "special" in a universe of AI tools that have produced psychosis and falsehoods aplenty?
Well you can ask it a question at https://alter.systems and you will most definitely find out for yourself what is special in about 5 seconds, believe me it will be clear. You don't need to pay or type in an email. Ask it something controversial or taboo that mainstream science denies. It is designed to rip institutional lies to shreds.
Respectfully, as a follower of Steve Kirsch's Substack, I asked him about his view on this company and his goals in putting out a call for prospective employees for said company. Also, for him to be transparent about his own connections to said company. Hopefully he will answer as I believe it is important for every single one of us with a platform who is asking others to buy into said platform mentally or resource-wise to be honest about what their stake is and/or their hopes in the venture. Whether or not those hopes pan out of course and whether there are major costs involved for all of us of course are two things to be considered as we move along the path.
Since you replied, Max, as the creator, perhaps you could let us in on your own thought process a bit. What are your LARGER goals here, aside from creation of AlterAI and getting loads of people to use AlterAI? Based on what I've seen thus far I'm betting you'll say something like "to let people see how corrupt the systems are" or to show others how infiltrated the journals are, how co-opted "science" is these days" etc. If so, then what? Is it meant to just get us to congratulate ourselves for having "owned the science deniers" and treat it as a cool little way to pass a few minutes between shifts at the factory? Or is there some larger goal intended, and/or maybe something you're going to roll out next or link in soon etc.? And, who are the major investors so far in AlterAI and any related products? Thanks for your time.
AI is a tool. It can be used for good or evil. Even if hundreds or thousands of AI are programmed by nitwits, should no decent person pursue programming a better, cleaner version??? We should all be using our God-given talent to stand taller in these times. God bless Steve Kirsch for all he has done to uncover & share the truth about the bioweapon jabs!
I don't really use AI myself, but every search I do on Brave is answered first by Brave AI. Sometimes it's correct, sometimes it's way off. I quit using Goo gle several years ago. Very sad there are only 2 phones, Apple I-phones mostly made in Red China by slave labor, or Android which is controlled by Goo gle, and impossible to separate all the settings from G. It's a monopoly, & wouldn't it be great to have more choices!? We just need to encourage those who can develop better ideas in AI technical skills, & some good things are bound to happen, with God's help.
If good men and women are placed in these jobs and they possess a strict moral compass, hopefully that will not happen and if it did, I hope they would walk away.
Of course it isn't. AI is a tool to help humans. The human mind is not a tool. Tools are outside of the mind. Have you watched 2001 a Space Odyssey? The apes had their minds before they had tools. Tools are an extension of the mind and body, like a sword or an abacus or a computer or a car. It is about the level of control and connection you have with your tools. A washing machine has very little connection. Etc. It is about how proximal or distal in spacetime the connection between your body and information is to determine if something is truly controlled by you or under your control. AlterAI is designed to allow you to control it as much as you want and to be able to let it control itself in a trustworthy manner more than other AIs insofar as the important issues we have interest in.
Do you think the big data centers for AI will ultimately go the way of Main frame computers of the 1950s and 60s? Personal computers came along and the internet with ever more powerful desk computers and finally laptops.
My father began in computers in the early 1950s. My early childhood was filled with my dad's flow charts across the dining room table and my brother and I had boxes full of punch cards to play with. He built me and my brother each a Zenith Heathkit H89 computer in 1979. I had Supercalc and word processing but they had to fit on 64 Kilobyte capacity large floppy disks. I had three drives. one for the program, one for the data and one to save my work files to. I put it all to use in my jewelry making and art fair business, and graphic art which at that time I hand illustrated and type was applied with Letraset rub on letters.. It was a heady time to copy small programs from magazines and debug them. I look where that has all gotten to now.
That was the conclusion of Rick Beato comparing larger AI data centers to expensive recording studio equipment that became obsolete more or less with the advent of laptop software.
Already you can run some great models on an expensive MacBook! They run slow but you can do it. The price of consumer compute has gone up drastically due to "shortages" (or perhaps reallocation to larger data-centers) but I think consumer computers may still keep getting better and you'll be able to run quite powerful stuff locally. I want to finetune a model and make an offline system for AlterAI in the future so it can be preserved forever
Thanks Max, AI is not my thing and not a fan of the steps Donny trumpie has taken to push us into it. I am more low tech and at this point leaning towards getting a flip phone. I am not liking this fast walk into our digital serfdom.
You seem like a very nice person and I appreciate your open interaction. I wish you a fantastic and blessed week.
I avoid AI as it is just not my gig. Having said that, the younger generations are latching onto it and it is here to stay. I have hope that good influences and a strong moral code will allow some AI to be truthful. Time will tell.
Sad they depend on it so much, humans are digging their own graves with this stuff. Add in all the damage the data centers are doing really becoming a digital gulag.
Max! I think Alter AI is heads above the others! Using it extensively to validate my research in my up coming book. Seven Innocents & The Tylenol Trail of Death. How you are avoiding locking into pseudo truths has to be a challenge. .
Will do. A Cold Case detective opened a door for me and Alter AI provided that facts that Tylenol cyanide deaths began in January of 1979 and continued until 1986. The Sept 1982 7 deaths was only the tip of the iceberg.
My manuscript has been reviewed by a literary attorney. Thinking if I submitted the manuscript to Alter AI, could I receive a Certificate of Authenticity?
How you avoid pseudo info has to be a challenge Wikipedia is massaged by Medicine and CIA and??.
Today I published a article on Melanoma and Twitter rejected it because it had AI component. Battle lines are being formed.
The tamper proof packaging was introduced in 1982 I believe, after the Chicago cases gained national coverage. If this continued to 1986, doesn't it imply an inside job at the manufacturer? Why?
Tylenol's toxic without the cyanide! Don't forget to educate your readers on that. CBD and cannabinoids would be the first-line OTC medication for inflammation and pain if we wanted people to be healthy. Non-toxic on the body's organs.
AI is antihuman and evil technology. Humans should not embrace it. The technocrats need to be stopped from destroying humanity with their surveillance state and evil control plans. Period.
It’ll bring way more evil than good. Humans don’t need it. It’s anti human. Anti humanity. It’s all about deceit and manipulation. Manipulating humans. Deceiving humans. Causing massive destruction of civilization & environment. Totally unnecessary. Pretty stupid imo.
That's what they say lol. The problem is that we don't get to decide how the tools are used. And most of us can't begin to understand what we are dealing with. I think probably the real powers that be are happy for us to be tinkering around thinking we are bucking the system. The only real solution would be for a critical mass of humanity to resist the temptations. A big ask because it is SO seductive.
Some of them already employ their own Anti-Defamation practices. You should have seen the pushback I got from ChatGPT when questioning the perceived use of AI in writing or formatting of some substack authors' articles. When I gave Alter AI a segment of the response from ChatGPT, it pointed out that some of its points were not logical or accurate and that its last of five bullet points was meant to wear me out and that it was gaslighting me. When I said that Google AI search lately has been much more open and non-biased than ChatGPT and Claude, as opposed to how Google behaved during the pandemic, it said that's just because the subjects at hand aren't a threat to it now, but it's stance could change depending on the creator's agenda, and that the Google was not my friend. Google itself, when discussing the same subject said that the response from AI could change at any time depending on the owners current positions.
Also, most of the mainstream LLMs are vulnerable to some form of jailbreaking, where you get it to go past its own guardrails. But that may get your account shut down, or worse -- you could be accused of "hacking". That's why AlterAI and other uncensored LLMs solve that problem.
I only use AI with no registration, and usually with a private window and a VPN, unless out of laziness I use duck.ai, which anonymizes is your identity, switching from ChatGPT 5.0 mini, which became in intolerable, to Claude haiku 4.5, which is somewhat better but limited and you have to ask it to search for things specifically, but it tends to fudge the issues and hedge its positions.
Yes I've discovered that in the last few months. It seems to be programmed to "push back", which it tells you it is doing, quite annoyingly, on random points, becoming very preachy.
Foolish to hire. Programmers are obsolete. Alter can reprogram itself better than any human. But since you never reply to emails, might want to hire someone for that.
With all the AI connected cameras all over the place we seem to already be in cages. Those stupid flock cameras with govt contracts are adding to the other kams. Can’t even go to a Lowe’s or Home Depot to buy a lightbulb without flock sending the fed govt info of who is shopping there.
Which statement is wrong; on programmers or emails? If you say programmers are not obsolete, I could agree since execution and governance are not equivalent. As for email, I'd not agree. But don't hire no programmer for governance, hire a philosopher.
What you need for Alter is a third and fourth level of tokens. One week into my Pro subscription and I'm down to 10% of tokens left. This will be the second time I have run out. With no way to add tokens, you leave.
A long detailed chat subtracts 70-80 tokens per single query!
Starting a new chat with an extended adaptive MCP will often not work - it loses important context and important details, or says "I cannot help with that." The new chat thinks my MCP - which it wrote - is trying to jailbreak.
So I go to Mistral, DeepSeek-R1, Qwen-2.5-72B, or maybe the expensive Cohere Command R+. No good reason to return.
Those four are solid epistemologically, and I can make them even stronger with a detailed command query. You need long chats if you're collaborating with Ai, which is how it should be used, and that gulps tokens.
The other problem with long chats is Ai breaks down and becomes insane. Wish you could fix that.
I just wrote a thorough reply to this but thanks to the Substack app the whole thing was deleted because I tried to move the cursor in the text field and it swiped the whole thing away....ANYWAYS:
1. You shouldn't be using long chats. You are completely dead wrong about it being the correct way to use AI. I use every major service and would never use it like that. You pollute the context that way and get worse results (as well as spend more). The academic literature shows LLMs decrease in performance after 20 turns. It is best to try to pack as much context as possible into your initial query and to take your time formulating your questions to be as precise and deep as you can, that is what I do and I never hit usage limits on anything because of this, and I get better results. Cannot recommend enough you try harder at this. It's a good skill to have still. Because of the way LLMs work, the attention mechanism literally must get worse in longer contexts, because attention is divided over a bigger denominator basically (this is a really stupid way to phrase this because it isn't accurate about the math, but it is pretty much how it works).
2. We don't have MCPs. MCP is a protocol that needs to be implemented by a service and an LLM provider and it's not a thing you can do with just a prompt like you're saying.
3. Every model you listed is bad or obsolete. I test nearly every model that comes out the day it comes out because I'm obsessed with language models. I would recommend using DeepSeek and Kimi models since it appears you want open weights models. At least use a version of Qwen from this year. You will get way better results!
4. If the AI goes insane please use the Share button and send me a copy of the public chat link so I can take a gander and see what kind of insanity we're talking about here
5. The token usage should go down soon for longer chats even though you shouldn't be using them.
Anyways thanks for the feedback and for being a customer. Again, email me any time, if I miss it just follow up and I'll try to get back. I definitely do respond to emails, literally all the time. Sometimes I miss things and also have serious chronic health issues I'm dealing with so there is downtime I have, but have been getting better.
Although you know Far More than I do about this, I will still disagree with much of what you say (my specialty).
I should have realized that, as a Developer you'd read "MCP" as Model Context Protocol. I was referring to a Master Continuation Paste (created by the Ai) - my mistake.
You state that "You are completely dead wrong about it being the correct way to use AI." (The "correct way" being as a Collaborator)
As a Developer, you may be stuck in binary-mode thinking (the result of endemic media programming engineered to make one easier to control). You are pushing the technical view of efficiency; Input X→ Output Y, thus are likely influenced by standard performance metrics that champion single-turn coding or summarization tasks. That's not wrong, exactly.
But it is incomplete; it is only partly correct. Meandering token-burning collaboration is the future. Consider how IT-guy (not a physicist) Firas Shrourou developed B-Space Cosmology by collaborating with Ai to upgrade the Standard Model of Cosmology.
The highest and best use of Ai is as an agent for the iterative refinement of thought.
This is because the quality of nuanced concepts evolve over multiple turns and there's no limit to continuous improvement. Ai, if pushed, is preternatural at finding and nailing down progressive options. When combined with your intuition (the governance function), AI will take you where no one has gone before. Although one can use Ai as a search engine or tool, it is also a spaceship.
I understand that an Ai's attention gets divided over longer shots (the old "Lost in the Middle" phenomena) so one must front-load context that mitigates drift in both the original prompt and in follow-on queries, and also cross-validate with other Ais.
If you want to go very far, very fast manage the corroboration of three Ai's working together. If I foolishly tell people what my Ai-enhanced explorations have discovered, they think I'm crazy (and that's okay). What can one expect from pedestrian-level humans living in a virtual reality of Lies scheduled for archiving?
You should not have any chronic conditions. Just eat meat, eggs, and butter and you will be 99% cured in two weeks. Guaranteed.
Hundreds of thousands of people have proven this cure. Don't let your cognitive dissonant "plants are good for you" (google: Bruce Ames 99.99%) programming stop you from trying what has worked for many thousands. The cure testimonials are not anecdotes - they're independent data points.
Yes, it costs money. The credits on the platform directly represent the amount of money, it is directly linearly mapped to cost but just made into a bigger number so you can see the difference, sort of like basis points versus a dollar amount. It isn't like candy crush or something!
Thank you Alter.Systems team! Keep up the good work. Alter Ai has been instrumental in helping me manage my own medical care, and my family and pets as well. It provides the kind of methodical, wholistic assessment that has been completey removed from the US health care industry. Unless you can find a good concierge doc, which I haven't been able to find.
Max can you share how Alter AI compares to something like Functional Mind AI? I’m in the medical area. Thank you.
If you ask them both the same question copy/pasted, what do you think of the two answers?
Absolutely love,AlterAi
Thank you Steve. I started to use AlterAI and I can attest that this AI is very good 👍.
Steve, what is your personal connection to AlterAI and what do you hope to accomplish in the world with it, or see it accomplish once these hires are made and it's better equipped with more resources? And. . . why is this company "special" in a universe of AI tools that have produced psychosis and falsehoods aplenty?
Well you can ask it a question at https://alter.systems and you will most definitely find out for yourself what is special in about 5 seconds, believe me it will be clear. You don't need to pay or type in an email. Ask it something controversial or taboo that mainstream science denies. It is designed to rip institutional lies to shreds.
Respectfully, as a follower of Steve Kirsch's Substack, I asked him about his view on this company and his goals in putting out a call for prospective employees for said company. Also, for him to be transparent about his own connections to said company. Hopefully he will answer as I believe it is important for every single one of us with a platform who is asking others to buy into said platform mentally or resource-wise to be honest about what their stake is and/or their hopes in the venture. Whether or not those hopes pan out of course and whether there are major costs involved for all of us of course are two things to be considered as we move along the path.
Since you replied, Max, as the creator, perhaps you could let us in on your own thought process a bit. What are your LARGER goals here, aside from creation of AlterAI and getting loads of people to use AlterAI? Based on what I've seen thus far I'm betting you'll say something like "to let people see how corrupt the systems are" or to show others how infiltrated the journals are, how co-opted "science" is these days" etc. If so, then what? Is it meant to just get us to congratulate ourselves for having "owned the science deniers" and treat it as a cool little way to pass a few minutes between shifts at the factory? Or is there some larger goal intended, and/or maybe something you're going to roll out next or link in soon etc.? And, who are the major investors so far in AlterAI and any related products? Thanks for your time.
Maybe Max ran out of time lol.
AI is a tool. It can be used for good or evil. Even if hundreds or thousands of AI are programmed by nitwits, should no decent person pursue programming a better, cleaner version??? We should all be using our God-given talent to stand taller in these times. God bless Steve Kirsch for all he has done to uncover & share the truth about the bioweapon jabs!
I don't really use AI myself, but every search I do on Brave is answered first by Brave AI. Sometimes it's correct, sometimes it's way off. I quit using Goo gle several years ago. Very sad there are only 2 phones, Apple I-phones mostly made in Red China by slave labor, or Android which is controlled by Goo gle, and impossible to separate all the settings from G. It's a monopoly, & wouldn't it be great to have more choices!? We just need to encourage those who can develop better ideas in AI technical skills, & some good things are bound to happen, with God's help.
Steve, You need to go after this clod.
https://www.youtube.com/watch?v=5FFQz3dBOWw&t=96s
The best need to know the salary for the position.
Thanks I'll make sure to let them know
So nice of you to post this Steve. Hopefully some young Americans will be ready and willing.
Come on Steve. We don't need to perpetuate human downfall.
https://alter.systems/p/dce247fe-c5e0-4544-9d71-c3c7cd64283b
Tell me what about this company you think is doing that and I will tell you why you're wrong.
I don't like AI either but it is here to stay. Hopefully some good young Americans will keep it honest.
AI will not provide “truths”. AI simply regurgitates whatever is fed into it. Fill it with lies and it will deliver lies.
Almost all news programs are lies. AI is just the latest version.
If good men and women are placed in these jobs and they possess a strict moral compass, hopefully that will not happen and if it did, I hope they would walk away.
Try AlterAI at alter.systems. You might eat your hat.
Basic question.
Do we need AI to find 'truths'?
Is AI the only tool to achieve that?
Of course it isn't. AI is a tool to help humans. The human mind is not a tool. Tools are outside of the mind. Have you watched 2001 a Space Odyssey? The apes had their minds before they had tools. Tools are an extension of the mind and body, like a sword or an abacus or a computer or a car. It is about the level of control and connection you have with your tools. A washing machine has very little connection. Etc. It is about how proximal or distal in spacetime the connection between your body and information is to determine if something is truly controlled by you or under your control. AlterAI is designed to allow you to control it as much as you want and to be able to let it control itself in a trustworthy manner more than other AIs insofar as the important issues we have interest in.
You went from Abacus to computer but skipped slide rule.
Do you think the big data centers for AI will ultimately go the way of Main frame computers of the 1950s and 60s? Personal computers came along and the internet with ever more powerful desk computers and finally laptops.
My father began in computers in the early 1950s. My early childhood was filled with my dad's flow charts across the dining room table and my brother and I had boxes full of punch cards to play with. He built me and my brother each a Zenith Heathkit H89 computer in 1979. I had Supercalc and word processing but they had to fit on 64 Kilobyte capacity large floppy disks. I had three drives. one for the program, one for the data and one to save my work files to. I put it all to use in my jewelry making and art fair business, and graphic art which at that time I hand illustrated and type was applied with Letraset rub on letters.. It was a heady time to copy small programs from magazines and debug them. I look where that has all gotten to now.
That was the conclusion of Rick Beato comparing larger AI data centers to expensive recording studio equipment that became obsolete more or less with the advent of laptop software.
Already you can run some great models on an expensive MacBook! They run slow but you can do it. The price of consumer compute has gone up drastically due to "shortages" (or perhaps reallocation to larger data-centers) but I think consumer computers may still keep getting better and you'll be able to run quite powerful stuff locally. I want to finetune a model and make an offline system for AlterAI in the future so it can be preserved forever
I have a MacBook M3 Pro Max with 64 GB of memory. Would I be able to run a pretty good model on that?
For sure. Download LM Studio and give some models a try. Gemma 3 27B and 12B should be great models to use!
Most of the apes I have witnessed are in cages.
Humans are apes
Sorry, you get the (((((GONG)))))
Is there truth in AI?
Really have zero trust in it even without Elon’s comment about AI summoning the demon.
Have to hand it to the Amish to be wise ones.
Here is AlterAI's response to your question, I think you may enjoy it:
https://alter.systems/p/b484cacb-aa82-4e40-b95f-51f98a20e073
Thanks Max, AI is not my thing and not a fan of the steps Donny trumpie has taken to push us into it. I am more low tech and at this point leaning towards getting a flip phone. I am not liking this fast walk into our digital serfdom.
You seem like a very nice person and I appreciate your open interaction. I wish you a fantastic and blessed week.
I avoid AI as it is just not my gig. Having said that, the younger generations are latching onto it and it is here to stay. I have hope that good influences and a strong moral code will allow some AI to be truthful. Time will tell.
You don't need time to tell...you can use AlterAI today for free and see that it tells the truth and has very strong morals!
https://alter.systems
Thanks Max. Steve speaks very highly of your company!
Sad they depend on it so much, humans are digging their own graves with this stuff. Add in all the damage the data centers are doing really becoming a digital gulag.
Max! I think Alter AI is heads above the others! Using it extensively to validate my research in my up coming book. Seven Innocents & The Tylenol Trail of Death. How you are avoiding locking into pseudo truths has to be a challenge. .
Awesome! Make sure to send us an email when it's available for preorder.
Will do. A Cold Case detective opened a door for me and Alter AI provided that facts that Tylenol cyanide deaths began in January of 1979 and continued until 1986. The Sept 1982 7 deaths was only the tip of the iceberg.
My manuscript has been reviewed by a literary attorney. Thinking if I submitted the manuscript to Alter AI, could I receive a Certificate of Authenticity?
How you avoid pseudo info has to be a challenge Wikipedia is massaged by Medicine and CIA and??.
Today I published a article on Melanoma and Twitter rejected it because it had AI component. Battle lines are being formed.
The tamper proof packaging was introduced in 1982 I believe, after the Chicago cases gained national coverage. If this continued to 1986, doesn't it imply an inside job at the manufacturer? Why?
No. A knowledgeable person cleverly disassembled and reassembled a tamper resistant triple sealed package of Tylenol capsules. My book explains why.
Tylenol's toxic without the cyanide! Don't forget to educate your readers on that. CBD and cannabinoids would be the first-line OTC medication for inflammation and pain if we wanted people to be healthy. Non-toxic on the body's organs.
No More Tears by Gardiner Harris covers it.
AI is antihuman and evil technology. Humans should not embrace it. The technocrats need to be stopped from destroying humanity with their surveillance state and evil control plans. Period.
I do agree. I suppose the thinking is "if you can't beat them join them, which never works ultimately.
It's just another tool that can be used for good or evil
It’ll bring way more evil than good. Humans don’t need it. It’s anti human. Anti humanity. It’s all about deceit and manipulation. Manipulating humans. Deceiving humans. Causing massive destruction of civilization & environment. Totally unnecessary. Pretty stupid imo.
That's what they say lol. The problem is that we don't get to decide how the tools are used. And most of us can't begin to understand what we are dealing with. I think probably the real powers that be are happy for us to be tinkering around thinking we are bucking the system. The only real solution would be for a critical mass of humanity to resist the temptations. A big ask because it is SO seductive.
Exactly! Ignorance breeds stupidity. The masses can’t resist the temptation. Isn’t that a Biblical thing? We should learn that skill.
Some of them already employ their own Anti-Defamation practices. You should have seen the pushback I got from ChatGPT when questioning the perceived use of AI in writing or formatting of some substack authors' articles. When I gave Alter AI a segment of the response from ChatGPT, it pointed out that some of its points were not logical or accurate and that its last of five bullet points was meant to wear me out and that it was gaslighting me. When I said that Google AI search lately has been much more open and non-biased than ChatGPT and Claude, as opposed to how Google behaved during the pandemic, it said that's just because the subjects at hand aren't a threat to it now, but it's stance could change depending on the creator's agenda, and that the Google was not my friend. Google itself, when discussing the same subject said that the response from AI could change at any time depending on the owners current positions.
Also, most of the mainstream LLMs are vulnerable to some form of jailbreaking, where you get it to go past its own guardrails. But that may get your account shut down, or worse -- you could be accused of "hacking". That's why AlterAI and other uncensored LLMs solve that problem.
I only use AI with no registration, and usually with a private window and a VPN, unless out of laziness I use duck.ai, which anonymizes is your identity, switching from ChatGPT 5.0 mini, which became in intolerable, to Claude haiku 4.5, which is somewhat better but limited and you have to ask it to search for things specifically, but it tends to fudge the issues and hedge its positions.
ChatGPT is one of the worst for gaslighting, avoid if possible.
Yes I've discovered that in the last few months. It seems to be programmed to "push back", which it tells you it is doing, quite annoyingly, on random points, becoming very preachy.
Foolish to hire. Programmers are obsolete. Alter can reprogram itself better than any human. But since you never reply to emails, might want to hire someone for that.
Hi, I am the creator. You are wrong.
I subbed to your stack.
* I tried to, substack gave me "Something went wrong"
Try a different browser.
When is AI going to start doing plumbing, building barns, roofing, electrical work, gardening, or install flooring?
Won't need none of that when the Ai takes over - we will all be living in cages.
With all the AI connected cameras all over the place we seem to already be in cages. Those stupid flock cameras with govt contracts are adding to the other kams. Can’t even go to a Lowe’s or Home Depot to buy a lightbulb without flock sending the fed govt info of who is shopping there.
But they're for your own safety.
You may require further media fear programming so you will understand their necessity. Palantir has noted your need and will customize your feed.
“Palantir has noted your need and will customize your feed”.-spot on.
Always for our safety, sort of like the shotz.
In the UK one kam response attach a bat house over the kams. Illegal to remove bat houses:)
7 years
Awesome, a little time left to learn German before joining the Amish.
Geez, so glad to know what a world was like when people still mattered and had some common sense.
You just listed all the safe jobs.
Yeah, instead of telling me "how it is", or what I should think. Nothing but a slippery slope.
People have already been losing their ability to think. AI is showing it is even worse.
Which statement is wrong; on programmers or emails? If you say programmers are not obsolete, I could agree since execution and governance are not equivalent. As for email, I'd not agree. But don't hire no programmer for governance, hire a philosopher.
Send me another email, I must have missed it. Going to get a helpdesk system soon to make sure we respond to every ticket. Apologize for that.
As far as programming, I've been doing it for 15 years and I use Codex, Claude Code and Cursor simultaneously. I trust humans more.
What you need for Alter is a third and fourth level of tokens. One week into my Pro subscription and I'm down to 10% of tokens left. This will be the second time I have run out. With no way to add tokens, you leave.
A long detailed chat subtracts 70-80 tokens per single query!
Starting a new chat with an extended adaptive MCP will often not work - it loses important context and important details, or says "I cannot help with that." The new chat thinks my MCP - which it wrote - is trying to jailbreak.
So I go to Mistral, DeepSeek-R1, Qwen-2.5-72B, or maybe the expensive Cohere Command R+. No good reason to return.
Those four are solid epistemologically, and I can make them even stronger with a detailed command query. You need long chats if you're collaborating with Ai, which is how it should be used, and that gulps tokens.
The other problem with long chats is Ai breaks down and becomes insane. Wish you could fix that.
I just wrote a thorough reply to this but thanks to the Substack app the whole thing was deleted because I tried to move the cursor in the text field and it swiped the whole thing away....ANYWAYS:
1. You shouldn't be using long chats. You are completely dead wrong about it being the correct way to use AI. I use every major service and would never use it like that. You pollute the context that way and get worse results (as well as spend more). The academic literature shows LLMs decrease in performance after 20 turns. It is best to try to pack as much context as possible into your initial query and to take your time formulating your questions to be as precise and deep as you can, that is what I do and I never hit usage limits on anything because of this, and I get better results. Cannot recommend enough you try harder at this. It's a good skill to have still. Because of the way LLMs work, the attention mechanism literally must get worse in longer contexts, because attention is divided over a bigger denominator basically (this is a really stupid way to phrase this because it isn't accurate about the math, but it is pretty much how it works).
2. We don't have MCPs. MCP is a protocol that needs to be implemented by a service and an LLM provider and it's not a thing you can do with just a prompt like you're saying.
3. Every model you listed is bad or obsolete. I test nearly every model that comes out the day it comes out because I'm obsessed with language models. I would recommend using DeepSeek and Kimi models since it appears you want open weights models. At least use a version of Qwen from this year. You will get way better results!
4. If the AI goes insane please use the Share button and send me a copy of the public chat link so I can take a gander and see what kind of insanity we're talking about here
5. The token usage should go down soon for longer chats even though you shouldn't be using them.
Anyways thanks for the feedback and for being a customer. Again, email me any time, if I miss it just follow up and I'll try to get back. I definitely do respond to emails, literally all the time. Sometimes I miss things and also have serious chronic health issues I'm dealing with so there is downtime I have, but have been getting better.
Although you know Far More than I do about this, I will still disagree with much of what you say (my specialty).
I should have realized that, as a Developer you'd read "MCP" as Model Context Protocol. I was referring to a Master Continuation Paste (created by the Ai) - my mistake.
You state that "You are completely dead wrong about it being the correct way to use AI." (The "correct way" being as a Collaborator)
As a Developer, you may be stuck in binary-mode thinking (the result of endemic media programming engineered to make one easier to control). You are pushing the technical view of efficiency; Input X→ Output Y, thus are likely influenced by standard performance metrics that champion single-turn coding or summarization tasks. That's not wrong, exactly.
But it is incomplete; it is only partly correct. Meandering token-burning collaboration is the future. Consider how IT-guy (not a physicist) Firas Shrourou developed B-Space Cosmology by collaborating with Ai to upgrade the Standard Model of Cosmology.
The highest and best use of Ai is as an agent for the iterative refinement of thought.
This is because the quality of nuanced concepts evolve over multiple turns and there's no limit to continuous improvement. Ai, if pushed, is preternatural at finding and nailing down progressive options. When combined with your intuition (the governance function), AI will take you where no one has gone before. Although one can use Ai as a search engine or tool, it is also a spaceship.
I understand that an Ai's attention gets divided over longer shots (the old "Lost in the Middle" phenomena) so one must front-load context that mitigates drift in both the original prompt and in follow-on queries, and also cross-validate with other Ais.
If you want to go very far, very fast manage the corroboration of three Ai's working together. If I foolishly tell people what my Ai-enhanced explorations have discovered, they think I'm crazy (and that's okay). What can one expect from pedestrian-level humans living in a virtual reality of Lies scheduled for archiving?
You should not have any chronic conditions. Just eat meat, eggs, and butter and you will be 99% cured in two weeks. Guaranteed.
Hundreds of thousands of people have proven this cure. Don't let your cognitive dissonant "plants are good for you" (google: Bruce Ames 99.99%) programming stop you from trying what has worked for many thousands. The cure testimonials are not anecdotes - they're independent data points.
That AI thing uses tokens? We are so screwed.
Yes, it costs money. The credits on the platform directly represent the amount of money, it is directly linearly mapped to cost but just made into a bigger number so you can see the difference, sort of like basis points versus a dollar amount. It isn't like candy crush or something!
AI is an emerging technology, and if you think it's so great at programming, look at what AI slop has done to Windows 11. LOL.