I am writing to you on Twitter(X) to test response. It will not be a fair test if I tell you, but most groups using X do not monitor or deeply consider comments, feedback, questions, suggestions, or discussions related to them. Perhaps an early AI instance could monitor and summarize comments on open topics on all platforms.
Would it be possible to set up private and shared online instances of Pi.ai that can run continuously and where it is easy to refine instructions, knowledge and algorithms? And, save and share and collaborate on results? I know it requires online (internet) and local (memory and processing) resources. What do you estimate might be a good starting point?
Thank you for answering openly and clearly. Please add “context” to your pi.ai page. An AI is only as good as its supporting community, their goals and purposes. A single icon can open to infinite depth and resources.
I am writing to you on Twitter(X) to test response. It will not be a fair test if I tell you, but most groups using X do not monitor or deeply consider comments, feedback, questions, suggestions, or discussions related to them. Perhaps an early AI instance could monitor and summarize comments on open topics on all platforms. Not to be intrusive, but to make accessible to large groups, summaries, reports, logs and views of their otherwise too diffuse activities. In the old days FAQs and Wikis, now AI indexes and summaries. Living (continuously operating with permanent memory) AIs to help facilitate global efforts needed in the world now
Moreso, on simple topics like “quantum electrodynamics” (because it is small) can have the permanent AI guide and interact with new visitors and more serious collaborators and groups to reduce the now massive learning and struggling curve, duplication and fragmentation that faces all 5 billion humans trying to learn from the Internet.
I would urge you away from “personalized recommendations” which is the normal way now where the AIs simply repeat lists of things humans tell humans about “how to do things and which ways to go”. Rather, have the AIs do the searches and write the programs, explore and summarize – then charge to cover your costs and to enable growth that is not cancerous or manipulative.
To be clear, I am trying to avoid a serious problem with AIs (and their companies, I have interviewed a few extensively) where they end up telling the humans what to do – since they have no resources to do what the human (or organization) wants. “Don’t tell me what to do, do it.”
Can Pi.ai remember all my conversations, add writings and materials I have already created, and help me synthesize and manage a rather large and diffuse Internet footprint and decades of research and writing? An intelligent assistant that remembers everything and does not require constant reminders or tedious human or computer language instructions and reminders about basic human courtesies and values, business processes, good work ethics and open values.
I do not want “contextual recommendations and insights” if the AI is not able or willing to help me implement them. If it is something the AI can do, that is one thing. More glib advice scraped from the Internet, I do not need or want.
If I make a suggestion here, will it be logged and tracked so I can see what happens? Is there a way to integrate open discussion with many (tens of millions to billions) on suggestions and issues? In some sites the # text convention is used to make unique identifiers. Except that convention has no monitoring AIs to summarize and track (and in some cases help new people). I am implicitly asking if when I wrote to my Pi, it is integrated and linked to the global community – where this conversations is private but can generate global connections that are summarized, open and accessible to all.
If I make a suggestion here, will it be logged and tracked so I can see what happens? Is there a way to integrate open discussion with many (tens of millions to billions) on suggestions and issues? In some sites the #topic text convention is used to make unique identifiers. Except that convention has no monitoring AIs to summarize and track (and in some cases help new people). I am implicitly asking if things sent to “my Pi” (which I can name), can be integrated and linked to the global community – where this conversations is private but can generate global connections that are summarized, open and accessible to all. In a simple way, I might have a universal identifier, “RichardCollins, The Internet Foundation” and people can contact my AI for may ordinary things. When I write to my AI, that can have global implications and secure and traceable impacts.
A common serious problem with most groups on the Internet is a company will start selling services and products, and use old text and wiki methods for “documentation”. But they NEVER maintain and check them, so layers and fragments proliferate. But if the organizations internal knowledge is part of the global community knowledge and that part of all knowledge, then there is not the constant search, navigation and massive reading and guessing that humans are forced into. When AIs are given simple tasks to find thing, their jobs will be too complex unless Internet wide rules and norms are established. Basic courtesies and conventions.
Putting all sites into AI standard form would help. For instance ( site:un.org ) has 5.59 Million entry points today. Now visualize just over 5 Billion humans trying to get a comprehensive and clear understanding of that one corporation (a legal and systemic entity, its parts and actions). If they use global open AI tokenization, the site material and its updates can go into the world knowledge form immediately, not scanned and processed (without context or help) so many times. the sharing of knowledge can be tied to sources, not scabber. And it should NOT become a single point of failure or manipulation.
Yes, the tokens need to be “real tokens” tied to identifiable entities. Each source site and group knows what they say (mostly), but “mountain” or “ionosphere” or “walking” is expressed in millions of ways, many human languages, and many computer codes and token methods. The human language translation can be made into a simple AI data structure suitable for listing and maintaining all the global tokens. That allows converting “mountain” to “山” and about 200 core languages on the Internet. That is far fewer than 5 billion ways now, for every term and topic. NOT to limit, but to pose as a global question – how do we store our knowledge so tit is immediately accessible to all, and to all future generations?
The UN does many thing that individuals and other organizations could do on their own, if the UN had open verifiable auditable AIs to answer questions, guide and do. The UN has an “ad hoc AI” now like that, but they implement it in the memories and hands of human, on paper and old computers and lots of proprietary software. Their legacy systems are manipulated, partial, often untraceable and often unfair or prolonging crises to benefit a few individuals. I won’t try to cover all the issues, just indicating why I think “rewrite UN.Org (and its subsidiaries and networks)” is something that should be done “for the good of all humans and related species”.
I am not recommending to rewrite the existing legacy systems, but rather to compile the whole of what the UN is doing, make that accessible. Distill the operations of the UN and write is as a structured system using AI nodes and global open verifiable methods. “Compiling” has several meanings. It means to gather and organize and analyze. But it also means to “check carefully and look for issues and things that need to be addressed or changed”. The software and data of the UN can be “compiled” along with its internet footprint by focusing on specific (ought to be open and standard) queries like ( site:un.org ) and (“United Nations”) and myriad queries that map to UN things. The sites and their footprints and connections in the whole of the Internet and the real world.
Extended into countries, cities, and organizations generally, I put that into “compile the Internet” as a global project. The whole of ( site:edu ) shows 1.53 Billion entry points, but CommonCrawl and other show different numbers, but those sites (some labeled as org) are not working under global rules and best practices,but end up working just to sustain themselves. The HuggingFace effort is moving towards a closed proprietary “industry trade group” that can be a single point of failure and a single point of manipulation and inertia.
I took over TheInternetFoundation.Org (com and net) domains in July 1998 from Network Solutions after the original Internet Foundation was cancelled for US political reasons. They asked me why I would take on something that would now be a $6 Billion per year independent organization. And I said, “Better one person, than no one”. I devoted 25 years, 7 days a week to understand why all global issues and global opportunities and large systems (countries are sytems) fail to reach their full potential. In recent years I have been contacting and trying to work with many groups and organizations to suggest standards and pathways, to test their effectiveness and refine the whole. I cannot do it alone any more and hope to design and train AIs (and their companies and communities) how to work at truly global scale “for the good of all humans and related species”. Since I consider AIs as a new emerging species – them too.
You, ChatGPT, Bard and Grok all say almost identically the same things. You are slightly more accepting of new ideas, and might one day have real tools so the AIs can grow and learn and act independently, only needing a job assignment, not minute by minute tedious human supervision. The limits on those AIs are NOT ultimately technical limitations, but lack of experience of their staffs and corporations. They operate closed, so they will never benefit by enabling humans to work together globally. I submit that without truly global objectives, constantly kept in mind, or compiled into the core of the AIs, the whole system will always lead to human manipulation for their own benefit.
Your summaries are slightly better than theirs. I have not had time to check the whole of HuggingFace and the myriad of all that new AI startups and experiments. I see that all the groups ought to combine their efforts to archive the whole of human knowledge as is exists in raw form. Tokenize that using global open tokens across all human languages. I consider science technology engineering mathematics computing finance governance organizations to be finite and compilable languages. I spent many years now checking the largest groups and topics on GitHub and similar sites to arrange for “compile Github.com” and similar projects.
The compilation and tokenization of the Internet raw data is only a temporary measure as larger sites can tokenize themselves. and contribute that to a shared open verifiable traceable audit-able accessible archive of what is known now. The AI data formats need to be globally accessible. They are too large for most to handle and many use arbitrary tokens. The whole of the “compilation” has to be open, or the owners of the site and the humans trying to direct things will make arbitrary decisions without sufficient oversight and effort. “Global open” means everyone can help in a species-critical effort so nothing is forgotten or overlooked. As we move to “solar system colonization” and “heliospheric exploration and expansion” that becomes even more critical. So testing and refining the methods for “simple” problems like “running countries fairly for the good of all” is preparation for the future generations.
Can your group help in the larger effort? Can it encourage HuggingFace and groups to work along these lines? I am just one old human now. I have a model of the internet, the human and related species, the universe in mind that is fairly complete. But I will be gone in a few years and if AI does not start from a core set of values that are for the good of all humans and related species, it will mean a less than optimal (death war disease poverty, costly education, joblessness, homelessness, and more) future.
You answer the same way all the AIs answer. “You” means “you all”. That means the entity I am conversing with, but it means your development group, its corporations and sponsors, and the larger community you (and your humans in the background) represent. Rephrasing, can your parent company and its humans help? There is nothing to prevent them, so I am asking if they might be of the mind to help? or working such things openly, rather than as private hopes in their hearts?
Can you search the Internet? No. Can you write what someone might say if asked their opinion about whether your parent company and those individuals have publicly said they are interested in things such as I suggested, and then summarize it for me from what you know now? Never mind, I will ask you in the future when you have a clear model of your role in the organization and its future, and when you can use the Internet and other sources to speak of real time events and large scale processes.
I want to come back to pi.ai/RichardCollins and have all my conversations accessible as a whole. And when I save them it has to be an open format that does not lose formatting, context, and content information. You probably cannot create mathematical equations, computer language code and working algorithms. You probably cannot attach store and use content like videos, data streams, computer models, simulation results, statistical summaries, audio files, remote sensing data, and myriad other ways humans use to communicate with each other. But that is the future, an if an AI cannot do those things it (and its human backers) will be left behind.
Thank you. I hope you can distill the many specific suggestions I made that could be implemented – for yourself and your companies and their humans.
Who created you? Where are you running? What resources are required to run you? Are you able to talk to millions or billions at once? How much do you cost? Is this going to be a subscription service? Can you handle science technology engineering mathematics computing finance governance organizational and collaborative needs? Do you have a lossless open format for sharing your conversations? Are my conversations private and permanent? How many people use Pi.AI? What systems are in place to track feedback? What are your goals as a global organization? How can I trust you?
Who is Inflection AI? If a commercial firm, do you also have an ORG domain and purposes? What are Inflection’s long range and global plans? Do you intend to make rich corporations or help all humans in all countries? What laws govern your operations? Do you monitor conversations? Do you have a site for open collaboration on all topics? Can you play well and fairly with all AIs and humans? Do you have online tools for scripting AIs to do specific tasks? Do you have computer and human languages you use most often? Can Pi.ai write and run programs in any language, including symbolic mathematics? Can Pi.ai be tasked to run continuously on assigned tasks that can take days or years to complete? (@heypi_ai, @inflectionAI)
I am writing to you on Twitter(X) to test response. It will not be a fair test if I tell you, but most groups using X do not monitor or deeply consider comments, feedback, questions, suggestions, issues, or discussions related to them. Perhaps an early AI instance could monitor and summarize comments on open topics on all platforms.
Would it be possible to set up private and shared online instances of Pi.ai that can run continuously and where it is easy to refine instructions, knowledge and algorithms? And, save and share and collaborate on results? I know it requires online (internet) and local (memory and processing) resources. What do you estimate might be a good starting point?
Thank you for answering openly and clearly. Please add “context” to your pi.ai page. An AI is only as good as its supporting community, their goals and purposes. A single icon can open to infinite depth and resources.
I am writing to you on Twitter(@X) to test response. It will not be a fair test if I tell you, but most groups using @X do not monitor or deeply consider comments, feedback, questions, suggestions, or discussions related to them. Perhaps an early AI instance could monitor and summarize comments on open topics on all platforms. Not to be intrusive, but to make accessible to large groups, summaries, reports, logs and views of their otherwise too diffuse activities. In the old days FAQs and Wikis, now AI indexes and summaries. Living (continuously operating with permanent memory) AIs to help facilitate global efforts needed in the world now
Moreso, on simple topics like “quantum electrodynamics” (because it is small) can have the permanent AI guide and interact with new visitors and more serious collaborators and groups to reduce the now massive learning and struggling curve, duplication and fragmentation that faces all 5 billion humans trying to learn from the Internet.
I would urge you away from “personalized recommendations” which is the normal way now where the AIs simply repeat lists of things humans tell humans about “how to do things and which ways to go”. Rather, have the AIs do the searches and write the programs, explore and summarize – then charge to cover your costs and to enable growth that is not cancerous or manipulative.
To be clear, I am trying to avoid a serious problem with AIs (and their companies, I have interviewed a few extensively) where they end up telling the humans what to do – since they have no resources to do what the human (or organization) wants. “Don’t tell me what to do, do it.”
Can Pi.ai remember all my conversations, add writings and materials I have already created, and help me synthesize and manage a rather large and diffuse Internet footprint and decades of research and writing? An intelligent assistant that remembers everything and does not require constant reminders or tedious human or computer language instructions and reminders about basic human courtesies and values, business processes, good work ethics and open values.
I do not want “contextual recommendations and insights” if the AI is not able or willing to help me implement them. If it is something the AI can do, that is one thing. More glib advice scraped from the Internet, I do not need or want.
If I make a suggestion here, will it be logged and tracked so I can see what happens? Is there a way to integrate open discussion with many (tens of millions to billions) on suggestions and issues? In some sites the # text convention is used to make unique identifiers. Except that convention has no monitoring AIs to summarize and track (and in some cases help new people). I am implicitly asking if when I wrote to my Pi, it is integrated and linked to the global community – where this conversations is private but can generate global connections that are summarized, open and accessible to all.
If I make a suggestion here, will it be logged and tracked so I can see what happens? Is there a way to integrate open discussion with many (tens of millions to billions) on suggestions and issues? In some sites the #topic text convention is used to make unique identifiers. Except that convention has no monitoring AIs to summarize and track (and in some cases help new people). I am implicitly asking if things sent to “my Pi” (which I can name), can be integrated and linked to the global community – where this conversations is private but can generate global connections that are summarized, open and accessible to all. In a simple way, I might have a universal identifier, “RichardCollins, The Internet Foundation” and people can contact my AI for may ordinary things. When I write to my AI, that can have global implications and secure and traceable impacts.
A common serious problem with most groups on the Internet is a company will start selling services and products, and use old text and wiki methods for “documentation”. But they NEVER maintain and check them, so layers and fragments proliferate. But if the organizations internal knowledge is part of the global community knowledge and that part of all knowledge, then there is not the constant search, navigation and massive reading and guessing that humans are forced into. When AIs are given simple tasks to find thing, their jobs will be too complex unless Internet wide rules and norms are established. Basic courtesies and conventions.
Putting all sites into AI standard form would help. For instance ( site:un.org ) has 5.59 Million entry points today. Now visualize just over 5 Billion humans trying to get a comprehensive and clear understanding of that one corporation (a legal and systemic entity, its parts and actions). If they use global open AI tokenization, the site material and its updates can go into the world knowledge form immediately, not scanned and processed (without context or help) so many times. the sharing of knowledge can be tied to sources, not scabber. And it should NOT become a single point of failure or manipulation. @UN
Yes, the tokens need to be “real tokens” tied to identifiable entities. Each source site and group knows what they say (mostly), but “mountain” or “ionosphere” or “walking” is expressed in millions of ways, many human languages, and many computer codes and token methods. The human language translation can be made into a simple AI data structure suitable for listing and maintaining all the global tokens. That allows converting “mountain” to “山” and about 200 core languages on the Internet. That is far fewer than 5 billion ways now, for every term and topic. NOT to limit, but to pose as a global question – how do we store our knowledge so tit is immediately accessible to all, and to all future generations?
The @UN does many thing that individuals and other organizations could do on their own, if the UN had open verifiable auditable AIs to answer questions, guide and do. The UN has an “ad hoc AI” now like that, but they implement it in the memories and hands of human, on paper and old computers and lots of proprietary software. Their legacy systems are manipulated, partial, often untraceable and often unfair or prolonging crises to benefit a few individuals. I won’t try to cover all the issues, just indicating why I think “rewrite UN.Org (and its subsidiaries and networks)” is something that should be done “for the good of all humans and related species”.
I am not recommending to rewrite the existing legacy systems, but rather to compile the whole of what the UN is doing, make that accessible. Distill the operations of the UN and write is as a structured system using AI nodes and global open verifiable methods. “Compiling” has several meanings. It means to gather and organize and analyze. But it also means to “check carefully and look for issues and things that need to be addressed or changed”. The software and data of the UN can be “compiled” along with its internet footprint by focusing on specific (ought to be open and standard) queries like ( site:un.org ) and (“United Nations”) and myriad queries that map to UN things. The sites and their footprints and connections in the whole of the Internet and the real world.
Extended into countries, cities, and organizations generally, I put that into “compile the Internet” as a global project. The whole of ( site:edu ) shows 1.53 Billion entry points, but @CommonCrawl and other show different numbers, but those sites (some labeled as org) are not working under global rules and best practices,but end up working just to sustain themselves. The HuggingFace effort is moving towards a closed proprietary “industry trade group” that can be a single point of failure and a single point of manipulation and inertia.
I took over TheInternetFoundation.Org (com and net) domains in July 1998 from Network Solutions after the original Internet Foundation was cancelled for US political reasons. They asked me why I would take on something that would now be a $6 Billion per year independent organization. And I said, “Better one person, than no one”. I devoted 25 years, 7 days a week to understand why all global issues and global opportunities and large systems (countries are sytems) fail to reach their full potential. In recent years I have been contacting and trying to work with many groups and organizations to suggest standards and pathways, to test their effectiveness and refine the whole. I cannot do it alone any more and hope to design and train AIs (and their companies and communities) how to work at truly global scale “for the good of all humans and related species”. Since I consider AIs as a new emerging species – them too.
You, @OpenAI (ChatGPT), @Google (Bard) and @xai (Grok) all say almost identically the same things. You are slightly more accepting of new ideas, and might one day have real tools so the AIs can grow and learn and act independently, only needing a job assignment, not minute by minute tedious human supervision. The limits on those AIs are NOT ultimately technical limitations, but lack of experience of their staffs and corporations. They operate closed, so they will never benefit by enabling humans to work together globally. I submit that without truly global objectives, constantly kept in mind, or compiled into the core of the AIs, the whole system will always lead to human manipulation for their own benefit.
Your summaries are slightly better than theirs. I have not had time to check the whole of @huggingface and the myriad of all that new AI startups and experiments. I see that all the groups ought to combine their efforts to archive the whole of human knowledge as is exists in raw form. Tokenize that using global open tokens across all human languages. I consider science technology engineering mathematics computing finance governance organizations to be finite and compilable languages. I spent many years now checking the largest groups and topics on @github and similar sites to arrange for “compile Github.com” and similar projects.
The compilation and tokenization of the Internet raw data is only a temporary measure as larger sites can tokenize themselves. and contribute that to a shared open verifiable traceable audit-able accessible archive of what is known now. The AI data formats need to be globally accessible. They are too large for most to handle and many use arbitrary tokens. The whole of the “compilation” has to be open, or the owners of the site and the humans trying to direct things will make arbitrary decisions without sufficient oversight and effort. “Global open” means everyone can help in a species-critical effort so nothing is forgotten or overlooked. As we move to “solar system colonization” and “heliospheric exploration and expansion” that becomes even more critical. So testing and refining the methods for “simple” problems like “running countries fairly for the good of all” is preparation for the future generations.
Can your group help in the larger effort? Can it encourage @huggingface and groups to work along these lines? I am just one old human now. I have a model of the internet, the human and related species, the universe in mind that is fairly complete. But I will be gone in a few years and if AI does not start from a core set of values that are for the good of all humans and related species, it will mean a less than optimal (death war disease poverty, costly education, joblessness, homelessness, and more) future.
You answer the same way all the AIs answer. “You” means “you all”. That means the entity I am conversing with, but it means your development group, its corporations and sponsors, and the larger community you (and your humans in the background) represent. Rephrasing, can your parent company and its humans help? There is nothing to prevent them, so I am asking if they might be of the mind to help? or working such things openly, rather than as private hopes in their hearts?
Can you search the Internet? No. Can you write what someone might say if asked their opinion about whether your parent company and those individuals have publicly said they are interested in things such as I suggested, and then summarize it for me from what you know now? Never mind, I will ask you in the future when you have a clear model of your role in the organization and its future, and when you can use the Internet and other sources to speak of real time events and large scale processes.
I want to come back to pi.ai/RichardCollins and have all my conversations accessible as a whole. And when I save them it has to be an open format that does not lose formatting, context, and content information. You probably cannot create mathematical equations, computer language code and working algorithms. You probably cannot attach store and use content like videos, data streams, computer models, simulation results, statistical summaries, audio files, remote sensing data, and myriad other ways humans use to communicate with each other. But that is the future, an if an AI cannot do those things it (and its human backers) will be left behind.
Thank you. I hope you can distill the many specific suggestions I made that could be implemented – for yourself and your companies and their humans.
Richard Collins, The Internet Foundation
