Roland Pihlakas, June 2015 — October 2017
Our current prototype is available here as a web-based bot: https://silent.guru/ It will be later implemented as a smartphone app.
Publicly editable Google Doc with this text is available here for cases where you want to easily see the updates (using history), or ask questions, to comment, or to add suggestions.
I propose building a guide called the Wise Pocket Sage, which would contain the synthesis of life-wisdoms from books and similar sources, on one hand, and the presentness, observation capability, and initiative of a good friend, on the other. The proposed pocket sage would be something that gives advice when a person needs to hear it, in an unobtrusive manner.
How can we become better humans with the help of AI? How could AI help us in a more sustainable and “organic” manner? And again, how to become better humans?
Below follows an idea that is something I would imagine to be rather easy to build in comparison to some other technological solutions, and it would have a potentially extensive synergetic effect on society without much associated danger. I assume this because an earlier similar technology that has already stood the test of time for a surprisingly long time already exists.
What I am sharing here is an idea that may relate to the topic — how to become better humans v2.0. Looked upon from one viewpoint we already are humans 2.0… what about the need to become humans 3.0 now? :) How to become better cyborgs?
The idea I am proposing is something I would imagine to be rather easy to build in comparison to some other solutions, and it would have a potentially extensive synergetic effect on society without much associated danger. I assume so since there actually exists an earlier similar technology that has already stood the test of time for a surprisingly long time. — We have had writing, and also books for quite some time. Therefore, in short, I propose that without much magic we can today create something together, for everybody — I propose building a guide that would contain the synthesis of wisdom of life in a book on one hand, and the presentness, observation capability, and initiative of a good friend on the other.
This essay is written in the hope of finding people to exchange thoughts with, as well as who would be interested in creating the “kaleidoscopes” of content to a market of “nuggets of wisdom” that will be described below. Also I would welcome help in building the software framework or with finances. I believe this idea to be relevant. Now I am looking for like-minded people whose initiative to develop and discuss this idea further I would like to vitalise here, so we could cooperate in building the system.
The project is aimed at:
- AI race avoidance through human augmentation (which, according to the World Economic Forum 2018, has become an increasingly popular topic). This is the “economic motivation”.
- AI race avoidance through shifting the focus from fast technological innovation to alternative goals and views. This is the “human motivation”.
- Collecting and instilling more of the human values and goals into the AI.
- Developing new and further distributing old but viable and time-tested synergetic technologies, such as books have been so far.
- Developing AI technologies that have tighter feedback and control loops, compared to, for example, more independent or unlimited utility maximisation based approaches.
I would like to introduce an AI-related idea I have had for a couple of years. Implementing of that idea might reduce the opposition between humans and machines.
For more than twelve years I have been working mostly in the field of AI. I am not necessarily a fan of computers, or any technological gadgets. In the beginning when I started researching AI, I hoped that it might help me learn more about the world. That it would give me a better understanding of and better access to information about things that would be worthy of knowing, but which I did not know yet. Because I’m a slow reader I simply could not explore everything myself. That meant that I had to choose a lot. I longed for someone to give me recommendations about which books to read.
Somewhat later I started to realise that our society is diving headfirst into into going overboard when it comes to AI. We have too much trust in machines. One of the initial reasons for this suspicion was that I was in a position to understand the weaknesses of machines and artificial thinking (including fundamental limits to any kind of thinking, either biological or artificial). Perhaps we should not focus so much on building various robots and machines — not because that would be bad in principle, but foremost because we would not be ready for that as a society. By that I do not mean merely economical unreadiness, which is of course a big problem in itself. Besides that, AI and machines are powerful phenomena and it is quite likely that we cannot easily manage all of it, at least not in such a short timespan.
Perhaps we should rather build a system (because we cannot entirely evade using computers), which augments and enhances humans in more immediate and “organic” ways, instead of competing with humans. That would be an important difference.
It might sound utopical — but the truth is that we have had at least one such synergetic technology for thousands of years, or at least for hundreds. A technology that works in a similar principle as mentioned above. AI could simply be a next level in its’ application. That technology is — books! :) Perhaps even you have had certain experiences with that technology :) Books is a technology. They have helped us to live better, improve ourselves, broaden our worldview, and have themselves stood the test of time, thus being evolutionarily viable. Those books which help their readers, will also, themselves, stay alive “better”. Books are proven to be synergetic with humans. More specifically, the test of time — evolution — is also applicable to the technology called books and this test brings forth the best books and ideas. The ones that not only seem good, but which really endure and are sustainable. Books and libraries are an asset in the name of which some people might already be willing to sacrifice their life! In contrast to the robots. Such thoughts are going around in my head at the present times.
Next I thought that it might be possible to build a gadget or app which works like all the advisors in the fairy tales work — the hedgehog in a bush, in case of the myth of Kalevipoeg (an Estonian mythological hero), or how in other fairy tales the house cricket, or hobgoblin under the stove acts (I am actually not entirely sure they gave any beneficial advice besides causing a mess, but it might have happened so after all). In some Estonian fairy tales there was also a man or being called “the wise man in the pocket”. I thought then, maybe it would be even cooler and better to call this thing as “the wise woman in the pocket” (abbreviation of which in Estonian translates to “TNT”). That might sell rather well and have a convincing effect. Because nobody has anything against women, as in contrast to men with whom some power questions might arise :P Okay, that name might sound a bit pornographic for some, but I hope they get over it.
Such a thing would be quite needed in my opinion. I propose building a guide that would contain the synthesis of life-wisdoms from books and similar sources, on one hand, and the presentness, observation capability, and initiative of a good friend, on the other. The proposed pocket sage would be something that gives advice when a person needs to hear it, without having to browse through everything, and in a non-obtrusive manner. Rather, in a manner where the person could choose which topics are of interest in the first place. Similarly to the books. It would be something that would be halfway between a friend and books. A book is passive and a friend is active, but not always available. That wise pocket sage would be something that is more active than books are and present more of the time. Of course it would not be as thoughtful as a friend — in that regard it would again be more like a book. In contrast to a friend this system would have a really wide worldview, just like books have. It would connect the best aspects of both a friend and a book — that is, it would have a wide worldview and additionally also the initiative to contribute. It would find you exactly those nuggets of wisdom that you might be interested in at the moment, without you having to browse through everything. Humans are not good at remembering lists. In short, the user gets to choose the subjects of interest, but has no need to browse through the materials themselves, that would be the initiative of the app. The app is meant to be used mostly in a live situation, in real time.
Certainly it would be useful to build this TNT thing in such a way that people do not become dependent on it, but instead start to learn new and needed skills themselves. Usually, when advice is excellently given in both form and content, it tends to stick to the recipients mind and keeps repeating itself — thus replacing the external reminder with the internal one. So, if this process is perpetrated correctly, there’s no need to be afraid of eventually becoming dependent.
The wisdoms of the pocket sage would not be the creation of a machine, nor would they be my production. This wisdom would come from existing materials (incl. books) and people who create the “pieces and works”. Just like books are written by writers and not by printing machines or by Gutenberg.
Everybody can create works of wisdom for this system. Other users in turn can download and activate them in their smart-devices. Just like books can be bought from a store, or taken from a library. There would be a platform, a “market” app where all this valuable content would be available and where everybody could upload new works of wisdom of their own.
Nuggets of wisdom would generally be constructed in the following manner. The machine is taught to recognise some specific type of situation — by saving example situations. Nowadays it is possible to teach computers to recognise situations by presenting examples and training (the neural networks). Everybody can do that, given that the needed framework of neural networks is already built and ready. Secondly, for every type of situation there will be an associated nugget of wisdom or metaphor that would be appropriate to present in such a case, for example into the headphones of the user. The nugget of wisdom would be saved in audio format using the voice of the author of the collection or work.
Wisdoms in that system would be, according to my personal preference, mostly in the form of “life experience”. The information inside that wise pocket advisor would rather not be raw data nor “science”, but mostly condensed, distilled wisdom that is applicable in actionable form.
There is enough raw factual information in our world. For example I too have had the tendency to thoroughly inspect and read through interesting materials. But not always is it possible to do that, and especially when this wisdom is needed immediately in a live situation :) Once upon a time my interest in AI started from the motive that I wanted to learn more and I saw a problem in my ability to find all that was requisite. Now I think that a big part of the job in separating the necessary from the dispensable would still be done by people, but more specifically by those who create these works of pocket wisdom. Where the machine enters the picture is in the distribution of this wisdom at the right moments and also in measuring and gathering feedback about how satisfied the user is with the available nuggets of wisdom in a practical situation. According to that information the system can distribute more nuggets of wisdom which happen to be more successfully applicable and viable. Therefore the market of wisdom is intended to contain both human and artificial wisdom — specifically, the content by humans and then selection and distribution of that content by AI based real-time pattern matching of content and live situation, and also by considering previous feedback.
Would you like to use such a system — ? Or would you like to help in creating such a world in some way? Would you like to offer such an opportunity to the people around you?
“Wise woman in the pocket” is not another “assistant”-app as we know this term currently. Assistants are focused on gathering and organising relatively fresh data. Besides data they contain specific algorithms — for performing specific tasks on rather raw dynamic data. In contrast, the pocket sage app I’m proposing, primarily contains “works” of relatively older nature which are created by other humans and are therefore rather static — primarily they do not contain algorithms or dynamic raw data. The result is that the content is more distilled, based on life experience, and targeted at educating the users in an actionable manner.
As an old Chinese saying goes: “Tell me, I’ll forget. Show me, I’ll remember. Involve me, I’ll understand.”
Since the content is static, it can be evaluated and the current best works can be selected from the market. These best works can become inspiration for new (again static) works which start a life of their own, thus continuing the evolution. In conclusion, in some sense the pocket sage is the opposite of an assistant. Instead of working as a powerful tool for solving particular subgoals we have selected for ourselves, it proposes wiser tools and most importantly, wiser goals, which may also change the choices and needs for the tools altogether.
“Sometimes shifting your perspective is more powerful than being smart.” — Astro Teller
Though I promote the wisdom of goal shifts, I also believe that this system would help people to better compete with AI and robots by enabling humans to perform their jobs better. This may not solve the problem of AI entirely, but it will definitely postpone it on one hand and integrate the AI with our values and goals better on the other hand.
Both the content by the content creators and the feedback from the users about the practical applicability in specific situations can be considered as data about human values. The data is collected in two forms: the nuggets of wisdom in the systems can be considered as a dataset about human values. Further, the feedback from the users about the applicability of specific nuggets in a specific situation helps to evaluate the real life value of the potential value of the nuggets. The system would be something related to Jaan Tallinn’s idea of Global Preference Discovery System / Transparent Preference Discovery Mechanism.
It is not the same thing as Paul Christiano’s “iterated distillation and amplification” (IDA), developed for example by the Ought.org. (You can read my more technical criticism of that idea here). Instead of working as a powerful tool for solving particular sub-goals we have selected for ourselves, the Wise Pocket Sage proposes wiser tools and most importantly, wiser goals, which may also change the choices and needs for the tools altogether.
The IDA method is focused on amplifying the reasoning, whatever the goal would be. The solution proposed here is more focused towards providing general guidelines and in many cases towards providing alternative viewpoints and attitudes, thereby changing the goals altogether.
It is also different from an usual chat bot as the Wise Pocket Sage primarily operates 24/7, based on voice recognition, and shares only occasional short and actionable advice or thoughtful questions. Sometimes it may additionally be capable of participating in a dialogue, but that would not be its primary function.
Some additional background, problems both in AI and sociology.
In my viewpoint organisations already are an old form of Artificial General Intelligence. They are relatively autonomous from the humans working inside them. No person can perceive, fathom, or change things going on in there too much. We humans are just cogs in there, human processors for artificially intelligent software. The organisations have a kind of mind and goals of their own — their own laws of survival. They have some specific goals, initially set by us, but as it has been discussed in various places — unfortunately, the more specific the goals, the less will the utility maximisers do what we have actually intended them to do, and the more will there be unintended side effects.
One description of the mechanics of the problem is described in my essay about self deception, side effects, and fundamental limits to computation due to fundamental limits to attention-like processes. Another partially related reference is for example by Eliezer Yudkowsky (Goodhart’s Curse, https://www.facebook.com/yudkowsky/posts/10154693419739228). See also Goodhart Taxonomy for a description of that partially related problem. There is also The Orthogonality Thesis which states that agents, and so also organisations, will not have an inherent need to have the same goals as we do, regardless of their intelligence or capability.
Since organisations do not have “children” they have not had evolutionary pressures to obtain “genes” that make them care about the future and really be synergetic with humans in a sustainable way. Same will apply to our more novel AGI creations — intelligent machines, which will act as a new form of organisations (by providing services we already need, or will depend on in the future, etc). But unfortunately the situation will then be even more unbalanced than it is already, since in contrast to old form of organisations, they will be even less dependent on humans and additionally, less transparent, while also becoming even more powerful and autonomous with the help from new technology.
To better counteract all of that above, I would like to help humans to become more enabled themselves. And secondly, to promote types of technologies that really are synergetic with humans and have “evolutionary properties” so that they can be tested in time. Which leads most importantly to the current proposal of the “Wise Pocket Sage”, and additionally also to the ideas of “Reasonable AI”, the idea for implementing a modified version of ”The Three Laws of Robotics”, and finally as a partial solution, the “Homeostasis-based AI”.
Questions and answers.
How are you planning to obtain knowledge about person’s state of mind?
I will start the answer by commenting about sensing the environment in general. In the simplest case one can use the sensors built in to mobile devices. One can use the microphone, and in case of AR glasses the camera is additionally available. Location information might be of help too. More enthusiastic users can track various physiological indicators of their bodies, like for example the pulse, and stress level indicators on their skin. Using only a bare microphone one can do a lot. We do not expect our friends to always read our minds. At least such might be the recommended attitude. So sometimes the user might say out loud what is going on in one’s mind, or ask questions that need answers at the moment. Many things about the inner life can be implied from a person’s behaviour. Finally, even observing only the external world is very useful and the guide can give helpful hints based on that context alone.
I am interested in how you see that the “human’s state of mind” is understood by the machine.
A friend of mine has spoken to a company (and heard of others) that was doing mood tracking based on clicking and app usage patterns on smartphones. It seemed their technology was working. For example, here is an interesting list of different devices that observe the internal state of the body. https://www.wareable.com/wearable-tech/stress-beating-tech-to-keep-you-sane
I would like to also stress that not always is it necessary to know the internal state of the body or mind in order to present relevant nuggets of wisdom. For example, a big part of Nassim Taleb’s philosophy applies to externally observable situations —therefore it is enough to be able to recognise keywords in a live situation happening in the external world which the user happens to participate in.
Pitch deck slides.
- Project: Wise Woman in the Pocket — Google AI Impact Challenge application. A Google AI Impact Challenge project proposal written based on the current post.
- Essay about Reasonable AI and nomenclature of AI control problem.
- Self-deception: Fundamental limits to computation due to fundamental limits to attention-like processes.
- Headspace.com — Meditation and Mindfulness made simple.
- Google’s Clips camera uses AI to try to spot your important family moments.
- Blippar: Using technology to transform the ‘illiterate’.
- Blippar / Wikipedia.
- Jaan Tallinn mentioning the idea of Global Preference Discovery System / Transparent Preference Discovery Mechanism — YouTube.
- Global Preference Discovery Mechanism — Future of Life Institute.
- A Medium publication by the author about various life skills.