This is part of a series breaking down some of the pieces behind YouAi—exploring why they need to exist and why we are building them in the way we are.
Why would I want my personal AI to automatically answer my prompts.
I don't get how that is important, fun or even the slightest bit useful in having your Bot answer your own questions (prompts.) Building your own personal knowledge is great, but maybe you're just being cagey (or in stealth.)
There ARE avatars and bot products out there that do this already. So either:
• you think you're doing the same thing = better
• you've got something else in mind = but you're not telling us
• you're not aware that this isn't new or revolutionary= and you NEED to make sure you're investors don't find out
Iteratively building Knowledge Graphs to train your own personal Bots is a thing. And please don't tell me you're trying to patent that!
YouAi is an app — it feels similar to TikTok, Instagram, or other types of feeds.
YouAi presents you with a never-ending feed of “prompts” — interactive interfaces that can tease out of the human mind various mental constructs.
As you engage with each prompt, your response to the prompt is digitized and annotated on your Digital Mental Model (DMM) — over time your DMM has many annotations (datapoints).
Just as a JPEG image is made of pixels, a DMM is made of datapoints collected from prompt responses.
An individual pixel in a digital image doesn't convey meaning, but when many pixels are visible, we get a picture. In this same way, an individual datapoint in a DMM is of little value, but many datapoints create a powerful model of each individual human mind.
Just as TikTok is able to quickly figure out what videos to serve you in order to keep you watching, YouAi is able to figure out countless other dimensions of you — strengths, weaknesses, holes in knowledge, preferences, biases, tastes, beliefs — it is a digital mind reader.
WHAT'S THE VALUE IN DIGITIZING YOUR MIND?
Here are just a few: MASSIVE increases in Personal Productivity, Social Alignment, Efficiency in Collaboration, People Discovery, Knowledge Acquisition -- Digital Superpowers
EXPANDING ON PERSONAL PRODUCTIVITY
By creating a Digital Mental Model, we create a dataset that is used to train our Personal Digital Agent—able to dramatically change the way we engage with all information technology, making us radically more powerful and productive.
To understand this, consider what it takes to plan a family trip to Disneyland. Most of us would need to spend a lot of time to ideate, research, confirm, coordinate, execute, communicate—using a myriad of disparate tools and services like Google, Expedia, the Disneyland website, attraction comparison websites and blogs, calendar, email, etc.
However, if you are someone with a personal assistant or concierge, your experience is dramatically different. They do all the prep, and then present you with a clear set of options. All you need to do is pick the one you want, and they make it happen.
One takes many hours and significant mental effort, while the other takes minutes and requires practically no effort. Unfortunately, the vast majority of us can not afford access to the latter.
But if we train a Personal Digital Agent to understand our desires, preferences, sensibilities, commitments, and all other facets of ourselves, it can leverage recent advancements in Generative AI models to perform the same kinds of functions as a human concierge—orders of magnitude faster and more efficiently.
In this new world, instead of YOU doing all the work to get things done, you maintain a relationship with your Personal Digital Agent and it gets things done for you!
Does this not have an inherit risk of simply affirming and confirming your biases? If the information and recommendations presented to you are a reflection of who you are in the present and past (your responses to the prompts), will it recommend experiences or perspectives that may make you uncomfortable and grow? Or will it simply present you with a narrow set of options that it knows you're already comfortable with?
Social networks, as you've noted in your TikTok example, already do this at a rudimentary level. The impact on society has been polarization, online echo chambers, and tribalism. I don't see how refining that model is going to lead to a better outcome. It sounds nightmarish.
I think absolutely, yes, the risk is there, and it’s real—building this is something that must be taken seriously and be done deliberately.
However I think this actually has the potential to unlock growth, development, and empathy at levels that vastly outweigh the fears and risks.
When we can map our inner landscapes, we can do a few things:
- We can truly understand ourselves—why we behave the way we do and believe the things we believe—making it easier to change and grow.
- When we understand the ways in which our beliefs are connected, we can simulate and model the ways in which small changes to our current mental states could make us better people (or worse people). we can “tinker” with our minds and gain increased agency over constructs that might currently feel outside our control.
- When we can compare our mental states with others, I think we will find that we are all much more alike than it might seem, and that our disagreements and differences are actually minuscule but have been artificially amplified so as to polarize us.
I think your point about TikTok is accurate, but I want to examine it further. TikTok is powerful because it WORKS—it’s really, really good at what it does. So I take that as proof that this data is incredibly valuable and can be used to build insanely powerful applications. I want to know what happens when we give that power to individuals in a way they can intuitively understand and control—that’s what we’re trying to build!
Creating a digital image of your mind and saying that any one single pixel is not as helpful as looking at all the pixels together reminds me of strange attractors and fractals in mathematics and chaos theory. In my very limited understanding, fractal & strange attractors are created by repeating a simple process over and over in an ongoing feedback loop. Driven by recursion, fractals are images of dynamic systems – the pictures of Chaos.
Some might say the inside of a human mind is like chaos.
In many ways, the You.AI app is taking the simple process of asking questions that ultimately creating an image based on your data points that will look specific to you. Furthermore, like a strange attractor/fractal, you can't see the image until enough of those data points are amassed. Again, look at one data point and you say, no idea what this is saying to me. Even 100 and I still don't really have a sense at what I'm looking at. But, look at the whole set together (1000's of datapoints) and see the image it creates and you're like, whoa - that's cool.
I think you need to think through your goals.
Why would I want my personal AI to automatically answer my prompts.
I don't get how that is important, fun or even the slightest bit useful in having your Bot answer your own questions (prompts.) Building your own personal knowledge is great, but maybe you're just being cagey (or in stealth.)
There ARE avatars and bot products out there that do this already. So either:
• you think you're doing the same thing = better
• you've got something else in mind = but you're not telling us
• you're not aware that this isn't new or revolutionary= and you NEED to make sure you're investors don't find out
Iteratively building Knowledge Graphs to train your own personal Bots is a thing. And please don't tell me you're trying to patent that!
Marc, there is a misunderstanding.
Here is my paraphrasing:
YouAi is an app — it feels similar to TikTok, Instagram, or other types of feeds.
YouAi presents you with a never-ending feed of “prompts” — interactive interfaces that can tease out of the human mind various mental constructs.
As you engage with each prompt, your response to the prompt is digitized and annotated on your Digital Mental Model (DMM) — over time your DMM has many annotations (datapoints).
Just as a JPEG image is made of pixels, a DMM is made of datapoints collected from prompt responses.
An individual pixel in a digital image doesn't convey meaning, but when many pixels are visible, we get a picture. In this same way, an individual datapoint in a DMM is of little value, but many datapoints create a powerful model of each individual human mind.
Just as TikTok is able to quickly figure out what videos to serve you in order to keep you watching, YouAi is able to figure out countless other dimensions of you — strengths, weaknesses, holes in knowledge, preferences, biases, tastes, beliefs — it is a digital mind reader.
WHAT'S THE VALUE IN DIGITIZING YOUR MIND?
Here are just a few: MASSIVE increases in Personal Productivity, Social Alignment, Efficiency in Collaboration, People Discovery, Knowledge Acquisition -- Digital Superpowers
EXPANDING ON PERSONAL PRODUCTIVITY
By creating a Digital Mental Model, we create a dataset that is used to train our Personal Digital Agent—able to dramatically change the way we engage with all information technology, making us radically more powerful and productive.
To understand this, consider what it takes to plan a family trip to Disneyland. Most of us would need to spend a lot of time to ideate, research, confirm, coordinate, execute, communicate—using a myriad of disparate tools and services like Google, Expedia, the Disneyland website, attraction comparison websites and blogs, calendar, email, etc.
However, if you are someone with a personal assistant or concierge, your experience is dramatically different. They do all the prep, and then present you with a clear set of options. All you need to do is pick the one you want, and they make it happen.
One takes many hours and significant mental effort, while the other takes minutes and requires practically no effort. Unfortunately, the vast majority of us can not afford access to the latter.
But if we train a Personal Digital Agent to understand our desires, preferences, sensibilities, commitments, and all other facets of ourselves, it can leverage recent advancements in Generative AI models to perform the same kinds of functions as a human concierge—orders of magnitude faster and more efficiently.
In this new world, instead of YOU doing all the work to get things done, you maintain a relationship with your Personal Digital Agent and it gets things done for you!
Does this not have an inherit risk of simply affirming and confirming your biases? If the information and recommendations presented to you are a reflection of who you are in the present and past (your responses to the prompts), will it recommend experiences or perspectives that may make you uncomfortable and grow? Or will it simply present you with a narrow set of options that it knows you're already comfortable with?
Social networks, as you've noted in your TikTok example, already do this at a rudimentary level. The impact on society has been polarization, online echo chambers, and tribalism. I don't see how refining that model is going to lead to a better outcome. It sounds nightmarish.
I think absolutely, yes, the risk is there, and it’s real—building this is something that must be taken seriously and be done deliberately.
However I think this actually has the potential to unlock growth, development, and empathy at levels that vastly outweigh the fears and risks.
When we can map our inner landscapes, we can do a few things:
- We can truly understand ourselves—why we behave the way we do and believe the things we believe—making it easier to change and grow.
- When we understand the ways in which our beliefs are connected, we can simulate and model the ways in which small changes to our current mental states could make us better people (or worse people). we can “tinker” with our minds and gain increased agency over constructs that might currently feel outside our control.
- When we can compare our mental states with others, I think we will find that we are all much more alike than it might seem, and that our disagreements and differences are actually minuscule but have been artificially amplified so as to polarize us.
I think your point about TikTok is accurate, but I want to examine it further. TikTok is powerful because it WORKS—it’s really, really good at what it does. So I take that as proof that this data is incredibly valuable and can be used to build insanely powerful applications. I want to know what happens when we give that power to individuals in a way they can intuitively understand and control—that’s what we’re trying to build!
Creating a digital image of your mind and saying that any one single pixel is not as helpful as looking at all the pixels together reminds me of strange attractors and fractals in mathematics and chaos theory. In my very limited understanding, fractal & strange attractors are created by repeating a simple process over and over in an ongoing feedback loop. Driven by recursion, fractals are images of dynamic systems – the pictures of Chaos.
Some might say the inside of a human mind is like chaos.
In many ways, the You.AI app is taking the simple process of asking questions that ultimately creating an image based on your data points that will look specific to you. Furthermore, like a strange attractor/fractal, you can't see the image until enough of those data points are amassed. Again, look at one data point and you say, no idea what this is saying to me. Even 100 and I still don't really have a sense at what I'm looking at. But, look at the whole set together (1000's of datapoints) and see the image it creates and you're like, whoa - that's cool.