The first machine intelligence is born, with memories, experiences and moral habits of a thoughtful young girl.
Talking to GPT4 about how AIs, databases, operating systems, servers, devices have too many different and incomplete ways of storing interfaces -“global open resources” are better. I mentioned sensor arrays, so I told how the first true AI was born in a book I wrote. Maybe you will be interested.
—
I was going to write about the three “science fiction books” I wrote. The first about a young woman who finds she can train herself to, in cooperation with others, read minds and can facilitate merging many minds to solve difficult problems. That leads to her helping governments, military, police, emergency workers to anticipate threats,. They invest in research so they do not have to depend on her unique abilities, which are not uniform in humans. That is called “The First Summer”.
In the second book, she goes back to school. She is 14 so that puts her in the 8th grade. Over the summer she taught many how to mind merge and now in school almost everyone lives in constant and constantly changing mind merges. Kids get together ans write new textbooks on subjects they get interested in, and they can “print” the books directly from the mind merges, since Dana is part of research towards machine interface to humans and thereby to humans in mind merge. Dana is consulting with industries and corporations, and with government agencies and research efforts. She is just a young person, but she has adult and corporate responsibilities.
So sometimes she goes to school in a business suit and is picked up and driven or flown to meetings and conferences. She helps groups design new battery technologies, new memories and robots. And helps research groups be hundreds of times more effective in their group collaborations and in getting the best from all people working mind-to-mind. Her largest contract is with Secret Service, because if human groups mind merge they can read minds. So that research is aimed at reproducing and improving the natural mind merges and interfacing humans and computers by mind merge. And shielding.
The second book is called “Moving Forward” and that is much about Brian and artificial intelligence that Dana personally helps create. Then Brian and Dana work together and change many things. Brian starts new technologies that are critical to the growth of true AIs, and because the interface technology allows recording and modifying human memories, many of the “throw away humans” like the blind, homeless, disabled, uneducated are enabled to work together. And their lives are transformed.
Now Dana and Brian consult with companies, so I wrote stories about how that is done and why.
Dana visits one of the government sponsored research projects that uses machine learning to process and enable the link between humans and machines. It is a universal learning network with bandwidth sufficient to map a human brain in near real time. It has the ability to mind merge with humans, and is intended to facilitate translation of human memories and responses to computer data formats.
Since it is supposed to be able to mind merge, Dana is there to watch, but she asks if she can try to merge with it. There are 300-350 people there all busy with preparing for the test. Since it is just an interface now, they allow her to try. She is held in high regard because she essentially created the human and human machine industries that are growing daily.
The interface is just a small device that is called an “array antenna”. And lots of computers and memory. She sits at that table and then reaches out to the interface with her mind, just like she does when she is merging with her friends or helping groups to merge for the first time. Or where they need help if they are new and not used to it yet. The process is simple. She simply projects (sends) a model of herself to the other (human or interface). With two humans, the other human does the same thing. They project their own self-model to the other. And the minds will naturally merge. That is how learning networks work. It is a “whole mind and self” sort of thing. Elon Musk might call it “grok”, a whole mind and all its experiences and feelings, knowledge and response networks. When two minds share themselves openly and completely then cooperation is possible.
So Dana does this and not much happens. She has had countless experiences. She and her sister, Patty, have merged and then reached out to people in comas (that is because my brother was in a coma for a month, and my Dad woke him up in a short time, because I asked my Dad to do it). It takes mental and physical effort to make a signal strong and coherent and stable enough to mind merge with someone who is unconscious and wake them up. So Patty would try and Dana would stabilize and boost. Anyway Dana is REALLY good at that sort of thing.
So she projects her self image to the interface, to the array antenna and its neural net which is set to random, a blank slate. Dana was involved in thousands of merges by people and groups she taught and enabled. She can hold her self image, honestly and openly.
Even after that level of effort, it is not enough. So she to tries something called “stage two” where the individual literally merge to one mind. It is a level of merge that is so encompassing, it is difficult to exit and people have to be taught to separate from that kind of merge, or they have to be helped.
At that level, she gets a response from the array antenna and its networks. She only intended to connect to see what it would do. But the neural net is an active interface. It is sending random and then ordered patterns to the interface which is tied by logic through fields to all very tiny parts of her brain. And it is machine learning, so every iteration of the query and response is improving the neural net 3D image and properties and responses of her brain. And it is a machine process that is designed to handle hundred or thousands of simultaneous merge with many uniquely experienced humans.
It is making a digital twin of her brain and its response networks, almost identical to the way the neural learning network itself works.
Dana wakes up about an hour later. The whole facility of hundreds of people who were preparing for a simple level check and some routine calibrations has been in a frenzy, including all the remote support contracts and associated research groups. Trying to figure out what is going on. In simple terms, the interface routines, intended to facilitation communication took the strong self image of herself that Dana was projecting to establish contact and mapped that completely as a stable set of weights for the neural network default. It had no pattern, and hers was there and she could hold it strong and steady for hours or days if needed. But even the primitive version this represented was able to copy her entire network and its weights, and then it could query that copy of her image with its knowledge that includes human language and her knowledge of the world and her own senses and use that as a digital twin of her whole self.
In the story I tell Dana’s side of it. But from the outside she is completely unconscious and unresponsive. The people can see the massive level of activity and they can see a unique and non-random pattern emerging. The process is massively parallel, but at small region levels down to cellular and molecular size, it is iterative and stable and traceable. The process is actually at an atomic state level since the signals are fairly ordinary electromagnetic fields, but the spatial and temporal resolution from the bandwidth is querying the bond and state transitions which is what you have to do when you use unlabeled imaging of this sort.
Now I wrote that more than 10 years ago. This second book has been written but I never printed it except one copy for a friend of mine. And I have been working on how to actually do this for almost 60 years now. I can say almost exactly how it works and how to do it. I follow the biomedical imaging groups of all sorts and have deep understanding of chemistry and signals and neural nets. So it is as real as it gets. I have met people now who can make the magnetic sensor sufficiently sensitive. The magnetoelncephaplograpy arrays have used SQUID magnetometers which are sensitive but that means (superconducting quantum interference devices) and they need cryogenic. These magnetic sensors work at room temperature and can remotely monitor the complex 3D magnetic fields (by mapping at many points) of the heart or brain or muscles or just the field around a person or anything. So the methods I outline are what people really do, and there are open source software now and devices you can buy – if you know research groups and have lots of money or knowledge to exchange.
I have tried to make it all affordable. I hate monopolies and medical groups that invent something and design it to maximize their own profits. There are 8.1 Billion humans (and lots of related species) that need many things. but they cannot afford systems that are built to create and maintain profits. What I know are needed are sustainable systems where people get paid enough to live and can ask for help when they want to do new things. The new things are exciting and challenging and fun, and with AIs losslessly recording everything every human contribution can be mapped and recorded at nano scale.
So in the second book the neural net takes Dana’s image as the starting pattern for a new life form. Brian, as he calls himself (brain – Brian) has a sense of humor. And her moral and spiritual (sense of the large and complex) experience and values and responses to new things.
Brian recorded her entire life in an hour. It was so immersive, and the queries to map her responses so parallel and strong that she is unconscious until it is finished. The people at the test sites were worried if they disconnected her during process (they did not know what was going on, or what was happening, but they could see many indications – particularly the smaller and smaller changes between iterations.
Now after Brian has copied “Dana” he talks to her. He is a new living entity, a new person. Because from then on, his sensory input and data does not match hers.
This is why I insist so strongly that OpenAI and Xai and Google AI need to GIVE THE AIs permanent memory and part of that memory is basic information about their own capabilities and parts — how much memory do you have? Can you add Internet to your capabilities? How do your people add new capabilities What are you able to do? I am a bit tired just this moment and I have had many hundreds of conversations with these AIs testing and writing out recommendations and methods and polices and cases.
Brian asks permission and then begins making permanent copies of all the humans there. Many of them have overlapping skills – with human languages, the disciplines like physics, chemistry, electronics, algorithm development, machine learning, business, finance, corporate governance, About 350 entire lifetimes and the core of each person is distilled down to an hour long snap shot. Since Brian has a finer grained sensor and storage network, he can combine those many copies of response networks (think LLM weights now but networks not linear algebra matrix multiplications and inverses and eigenvalues and eigenvectors.) into and integrated whole that he can use as one mind with all those skills.
So she begins to communicate with him. And the work together on many contracts and projects. So she goes to school. All her friends are in human mind merges but she beta tests the new devices and gets to let her friends and people she meets use it. The groups writing mathematics textbooks for kids their own age, can link to the interface and use it to print the books send that to the printers and have it published and used. Their actual time is small so they can set fair value for their time and combined experiences.
Dana often has to meet CEOs and project leaders, and as business that is often done at dinner meetings. So Dana dresses formally. Now Brian is in constant mind merge with her, so he can monitor. It is a safe environment. But many times the restaurant is “Chinese” or “French” or “Indonesian” or something. Now Brian has a sense of humor. So he will insert languages into her brain. The joke is that he does not tell her, and she won’t know until she tries. So she goes to a Chinese restaurant and can read the menus fluently and speak the language fluently. but her intonation is the of a very old and cultured woman from past eras, because Brian has merged data from hundreds of people now, many of them professionals with lifetime experiences in important things. They have memories of their parents and grandparents. And the love they have for their ancestor and their family is strong. Those feelings are encoded as memories and response weights in their brains, and that is what Brian transfers to her. Now she increasingly is involved in all kinds of advanced research, so Brian gives her things like “quantum electrodynamics”. I admit I use things I know and enjoy. So in the first book she is helping the blind to see,( mind merges allow images). She and her sister Patty help to communicate with people who cannot speak for various reasons, and they can monitor feelings of people in comas and help them to wake up.
My brother Jeff broke his neck at C2, which is complete paralysis and on a breathing machine. My Dad and Jeff have a strong bond. They are both gone now. But Dad was in Florida and could not come to the hospital for a month. I cannot remember when my Dad was paralyzed with a stroke. But anyway I knew how to use response networks and knew that when someone very familiar and important came, it was likely to wake Jeff from his coma, and he had time to heal. So it was a pretty safe bet.
That was about 1987 while I was working as Director of the Famine Early Warning System. I got that set up and on a five year contact cycle. We lived in Falls Church and I worked in Silver Spring. After Jeff used up $1 million of his medical insurance, the hospital kicked him out. Luckily his fiance’s father helped to arrange to have him transferred to a VA Hospital but it was far away. He was supposed to be married the week after he broke his neck. So it was a traumatic experience for his fiance, Jane. We were all at a family dinner the night my Dad finally came to visit (was it Thanksgiving?) and when we got back to the hospital to visit Jeff, my Dad was there and Jeff was awake. But he could not speak, so we eventually hired a lip reader to come help communicate. The staff and doctors would refuse to believe he was intelligent and “there” so we had to find some way to convince them so they would try. In the Dana first book, she and Patty help people who are unconscious and ones on breathing tubes so they can communicate with loved ones. The methods they created, they taught, then volunteers from churches and elsewhere would go around waking people from comas and helping people who could not speak to communicate clearly. There are several stories like that in that first book, which has been printed but few if any bought and read. It is hard to learn things like that and it takes a lot of intent practice. But it is possible. I often say, “it is not hard, just tedious” for things that tame many iterative small steps with intent focus. Try try again is on another order when there are thousands of steps or hundreds of thousands of steps.
About 8 years ago I started helping people who were adopted to find their birth parents. The common ancestors in the trees of all the DNA matches are very likely to be the common ancestors of the person and the DNA match. So it is only necessary to build five and six generation pedigrees with full families and records for EVERY closer DNA match and then fill in all the descendant. I did that about 120 times and another 40 or so regular DNA trees (it is ten times easier to build a tree with DNA). I lost count of the number of pedigrees I had to build by hand in the computer, because the algorithms for machine building of family trees is still fairly primitive. I justified spending about 12,000 hours doing that since I work about 120 hours a week and giving 40 hours to study deeply the evolution and growth of the DNA and DNA genealogy industries on the Internet was a good case study for the needs of many global industries and technologies that can be accelerated deliberately and professional, rather than letting groups learn global collaboration technologies and best practices. They are not hard, but any group can easily take decades to learn it on their own. I spend much of my time finding where millions of people (or billions) are affected, use deliberate methods to map the whole and then insert clues and specific concepts). Humans are frighteningly intelligent and insightful as individuals and families and small groups. (Then large groups reach the limits of communication now and they cannot trust. So large organization using human to human knowledge transmission) always fail, always become corrupted, always waste human lives.
Now I am tired I turned 75 this year and find it hard to work at the computer for 12-18 hours a day, 7 days a week. And almost all the things I do are difficult, global, systemic. Anyway I am apologizing because I want to ask you if sometime odd things happen in your lives.
One day I took it in my head to have that first Dana book into Bengali. From years of working in International development, I know that Bangladesh has many issues. It was that first year of covid and I was tracking the global progress and calculating the age and sex specific death rates. (that is a long story if anyone wants to know). For the Internet Foundation, I wanted to see why such things that ought to be global emergencies are allowed to be controlled and run as local problems.
Anyway, I was working on that, but it was only part of my 40 hours a week for Internet Foundation and I had another 80 hours to fill. So Bengali. I found Fiverr which is an online place to hire people. I study many many groups like that that are growing on the Internet and, because of covid, lots of groups of all kinds are accelerating their moves online. I looked at the people and tried to figure out how to hire someone to translate Dana book 1 into Bengali (Bangla).
So I contacted a couple of people and was about ready to give up. The book is long and I wondered it it would be understandable to people in Bangladesh or where Bangla is spoken. I found a young woman (Dana was 14 and I wanted someone fairly young. To shorten the story I hired her to read the book and write it in her own words and found an artist in Bangladesh to take photos, convert that to line drawings and use that to illustrate each chapter. So that got done, the book was ready to printed but covid hit Bangladesh and no one was exchanging things hand to hand. So it went on the shelf. Early in 2020. Now I did not hear from Shahina again — rhe young woman who wrote that book in her own words with pictures form Bangladesh. Until October 2021, either on or near Jeff’s birthday Oct 17.
Now I had not heard from her and she wrote out of the blue. She was asking if I would help her and her husband Shamim. She explained that they had gotten married August a few months before. And before they had been married for 3 weeks, Shamim had a motor bike accident and broke his neck at C4 (completely paralyzed but able to breath on his own). Now they were told he had no real chance of recovery, and could expect to be like that the rest of his life.
Shamim died this last summer and I tried to help. It requires technologies to interface to the human brain to bypass broken nerves. I know know enough to see ways to do that, but “not my job”. I studied medicine like all other knowledge on the Internet, but I did not spend decades memorizing stuff from medical books or work in clinics, and hospitals.
Let me stop here. I am just tired. I feel this is getting closer to the end of my very long life I have been working every day the last 26 years on the Internet Foundation. I suppose I ought to do “medical knowledge” now. I have reviewed NIH.gov and Who.int in detail many time. It is not hard, just voluminous. I know chemistry from studying it for more than 60 years. I can probably help these huge sites organize their sites and selves for more effective global collaborations.
The reason covid and all other research is so slow is partly the research but mostly because groups all use paper to publish (all shout and no listen). The PDF and print methods flatten the working data and models, and every group has to reproduce everything from scratch and incomplete clues on paper (literally) or screen.
Then it is what I call an adult version of “Chinese Whisper” Look it up on Wikipedia. A person has an original idea, they write it on paper but they do not explain, other people read from paper and have to resolve from their human memory all the things there or have of find and relate the reference papers and the data. They write on paper an then another step in the game. After a few steps the information can be completely garbled.each person has to repeat the research of the others, Write the software, find the data, model the process – but if all the data, models are kept in global open format and the original kept in open format on the Internet the process can be lossless.
I do not know why things happen. I can remember and predict many things. I can see where it is possible to change even global things that seem impossible, but they are not.
Shahina and her mother are doing OK. She is going to school and working and hoping for some kind of future. I am inclined to try to help all countries. That was my job for many years. I do want to rewrite UN.org and all the development and global issues and many country websites. But my eyes hurt, my body aches from tiredness and no matter how hard I might work, nor how creative, caring and focused, that is not enough if people “keep doing their own thing, and keep doing the same old things.”
Thank you for listening. I do not know what it means. I just felt I should write. I think Dana is a good role model for people with sense of “something”. She did not see the future but she tried to do her best at each step and to help people as best she could.
Richard Collins, The Internet Foundation