Artificial Reality


The next step Feb 3, 2016

Are we creating a next step in the many social revolutions that technology has brought so far? Our bodies are made of collaborating living cells. Somewhere in evolution they decided to work together and specialize. The whole being greater than the sum of its parts, a concept known since Aristotle.

Communication is common in even lower lifeforms. However in the case of people language abilities developed into a sophisticated tool to connect and develop minds. Sharing memories and experiences through generations has lead to societies as we know it. Young humans are raised - their minds shaped and developed to comprehend ever more complex concepts. If they are lucky enough to be born in the right environment that is.Very few people ever will reach their full potential. Most people will be stuck in a struggle for basic needs. Besides that, it’s unclear what the full potential of the mind really is.

In various fields of science cooperation has resulted in astonishing insights and innovative technology. Eventually even providing intelligent communication systems that are expected to further evolve.

Into what is however the question? Technology evolves but people adapt. Consider the way we can use the web nowadays. Instead of having to remember every detail or carry around books, almost everything can be looked up. Information at your fingertips. As long as you know what to look for and how to recognize the relevant source. Search engines are increasingly improving making this easier and easier. Add in speech recognition (and synthesis) and what do we have? An internet full of knowledge. Databases full of user data. Speech and text based communication. All but a central processing unit to make sense of it all, so we can just ask questions and receive intelligent, clear answers and assistance.

With the technology of today it's partly possible. Large amounts of data can be decomposed to singular values. It's all about statistics, determining the most probable answer to a given situation. Incredible results can be achieved with self learning systems. Once feedback can be processed into improved answers a powerful loop is created.

Still we're missing something. How about actually understanding us and our reality?

The reality bubble

What we experience is our reality. The outside world gives us input and we give it meaning.

How we perceive reality can differ, but on many things we agree. We all experience the basic laws of nature around us through our senses. We can recognize consistent patterns and through communication compare our experiences.
In our minds we all carry a model of how the world looks and works. It might not be a complete model - but it serves our daily needs. People are always looking for consistencies and patterns to make sense of phenomena. New input needs to be placed in the existing model we have. It has to make sense.

That defines the way an intelligent system should work that could relate to people. It should be able to fit new information into a consistent model. Sometimes it's not that sure, on other occasions it just fits right in.

Within computer science a model for describing the world that consists of a set of types, properties, and relationship types is called ontology. A practical application already in use is the semantic web where metadata describes the relationship between data. Mostly to inform web spiders about anything from the author of an article, his place of birth to the price of a certain offered product.

Connecting our minds

When you connect computers to each other they can perform tasks in a parallel or serial way.

Serial means you can consider it to be one big computer with the sum of the processing power of the connected computers. It would perform the given task relatively fast.

Parallel computers would require the original task broken into little parts. Each part handled by a single computer. One of more computers would have manage these partial tasks and compile the partial outputs into a single output.

Seems like a lot of overhead. But just like with serial switched lights, the whole chain is broken when one light breaks. The broken light in a parallel switched lighting wouldn't affect the others - except they have to handle the same electrical current amongst fewer working lights.

The human brain is just like a single computer in that way that it's a unit capable of individual processing. Societies are formed by organization in a parallel way. Ideally people perform a task in society while others manage the tasks. The benefits are that larger tasks can be accomplished than a single individual could ever handle. This collaboration strategy has worked well for many species.

Serial switching of humans seems impossible anyway. You can't put two averagely gifted people together and fuse their mind into one brilliant one. What you can do is educate other people, learn from them, help them, let them help you. Thus adding to the organization of parallel performed jobs.

With this all in mind it seems plausible to say it's our destiny to organize ourselves even further. Technology is now in the driver's seat of this development. Which is only natural as it constantly improves the way we can communicate.


Big system is watching you

The dystopian novel Nineteen Eighty-Four - published in 1949 - is widely referred to whenever personal rights make way for reasons that concern perceived common interests.

Rather than the future it pinpointed the mechanics of the society that Joseph Stalin had created in the Soviet Union and that was at that very time being enforced on Eastern European countries.

Deception, secret surveillance and manipulation in service of totalitarian oligarchs. Nothing new since the dawn of ages of kings. Actually every society form seems to eventually result in an oligarchy as was first described by the German sociologist Robert Michels in 1911 as the Iron law of oligarchy.


The dilemma of personal freedom versus state interests as in the Cold War situation described by Orwell has made way for a larger stake: the common interest, people’s personal safety against crime and terror, outweighing personal rights by popular demand.

Everybody can agree that the boundaries of personal freedom are reached when someone becomes a safety threat to other people. Most can agree upon this boundary being set when it comes to one's own safety. But a threat to larger collectives (nations, religions, languages, families) is something that can be argued upon. To share opinions in a peaceful way is one of the core values of democracy.

Biometric identity verification has come a long way: iris scanning, fingerprints, even smell detection have improved significantly. With this in mind we should be able to rely on verification tools in the future. In combination with smarter information technology there is an opportunity to guarantee privacy and make society safer at the same time. Certain aspects of the private domain are irrelevant or too trivial to be of assistance in a possible criminal investigation. Other aspects might only be of interest in aggregated form for general statistics. However if in the public domain everyone is identifiable. Incidents can be captured on video by recognizing them - meaning you don't have to store every recording.

Constant surveillance can be a good thing as long as it's really for the good of the people - including yourself. And nobody can abuse this information. Only in certain cases some information is shared. Suppose you wanted to hurt someone - or yourself. Better to prevent this than to have to be sorry later.

Children don't always grow up to be sane and conscientious adults. According to research (NIMH, 2007) as much as 9.1% of the population could be classified as having a personality disorder - not mentioning other mental disorders. The ones that suffer are usually the surrounders. A lot of crime and violence can attributed to untreated personality disorders, especially antisocial disorders. Suppose personality disorders are noticed in a early development stage? Intervention in such cases will result in fewer crime as research shows a great deal of disorders being involved.

Assisted development of children can help develop them reach their full potential. Having a fully developed conscience is necessary to participate in society. Developing every single child to this level - and taking care of those who fail to socially adjust - are primary tasks for the kind of society we want for our own children: free and safe.

Some might argue that it sounds a bit totalitarian. Don't forget: we don't have a choice. The possibilities are there and they will be used or abused. Computers don't abuse people. So you have to make absolutely sure people can't abuse computers. That's why it might be a good idea to consider by what rules we want to deal with the inevitable.

Intelligence and knowledge

The word ‘flower’ usually refers to the blossom of a plant. As people we understand the concept of a flower. They come in many varieties and colours. Plants grow them for reproduction purposes. People give them to other people on certain occasions.

Now how would you explain all of this to somebody - or something - who knew nothing about the concepts we humans share? Mention the colour and you'd have explain that light in different colours is perceived by your visual senses. If you're educated you might be able to explain that we can see light in the visible spectrum and can't detect UV and IR. You'd might explain further and call it an electromagnetic wave - one of the fundamental forces of nature.
When you start describing an object or a concept, through its properties and relationships it connects to the rest of the universe as you know it and understand it. You'd end up describing your mental mapping - your reality bubble. The boundary of the unknown being a greyish field full of speculations.

Intelligence is defined in many ways, often referring to certain aspects, mostly based on the study of ourselves and animals.

According to Wikipedia:
Intelligence stands for one's capacity for logic, abstract thought, understanding, self-awareness, communication, learning, emotional knowledge, memory, planning, creativity and problem solving.
About Artificial Intelligence, same article:
Among the traits that researchers hope machines will exhibit are reasoning, knowledge, planning, learning, communication, perception, and the ability to move and manipulate objects.
Perception and manipulation of objects were not previously mentioned as aspects of intelligence.
Interesting enough researchers don’t seem to hope to hope for aspects like abstract thought, understanding, self-awareness and emotional knowledge.
Why not? Well, why would be a better question. Computers serve a purpose. That’s why perception and the manipulation of objects come in handy. Also aspects like emotional intelligence are so complicated and typical human that they would require a thorough understanding of our communication structure. Any system so sophisticated would not only be intelligent but conscious of itself. Once it starts working on its reality model it would have to include its own existence - it might even be the starting point of the model.

Subjective interpretation

Our questions and demands relate to ourselves. Demanding meaningful actions requires a context that extends our mutual understanding of reality.

It's also about understanding you. Knowing where you are, what your schedule is, what your plans are and preferences. Remembering previous interactions.

Let's see how we would have to answer the question 'How is the weather today'? Understanding of the semantics wouldn't be necessary - search engines would treat this kind of FAQ in a more statistical way. The concept of weather automatically raises the needed location. A subtract of localized weather forecasts would be a correct answer for a search engine.

But let's go a step further and personalize this. In that case a correct answer might be: "Great weather to go fishing this morning. Don't wait too long, it might start raining later this afternoon". This would require a much more sophisticated system that comprehends the concept of weather and what it means for your activities.


Meet your Personal Assistant

What do we want foremost besides basic needs and good health? The security and safety of their loved ones would rank high. A future in which children can develop their talents and lead a happy and fulfilling life.

Soon to come self driving cars will save the lives of many people. Human drivers make mistakes and neglect the rules. They need more space on the road and waste energy - it makes sense to replace them by autopilots but driving is too much fun for many.

Now take one more step and imagine a world in which robots can perform tasks in the household. And communicate effectively with you. Such a sophisticated service would have to connect through a fast wireless connection to a central system.

Your robot is just an household appliance so you might as well talk to the central system and let that handle the appliances. Any access point would do, so in your own home you can just talk to the system anywhere. Such a service would require private processing and data storage as well as the use of shared services.

As proposed these services could include a shared reality bubble: a organized long term memory that is accessed by a processing system Comparable to our short time memory this receives input from its 'senses' and relates this to its long term memory to find the right patterns for a reply or action. Neither the knowledge base nor the processing would would reside on any specific hardware. It could be spread around in a distributed manner with central mapping and routing services connecting them - just like DNS for websites.

Wouldn't it be great to have an intelligent assistant who can manage all your online accounts for you? Who can instruct the household appliances and have you picked up at a party?

Your assistant is a close friend of all other personal assistants of other people. Directly representing you and your interests without any judgment or opinions. Data about your likings, viewpoints and principles enable an almost real time democracy.

Eventually so much of you is known that in a way you have become virtual. Your personal assistant resembles you and will survive you. One could imagine it being of comfort for the loved ones of a deceased person.


The Shared Reality Model

We all carry a model of reality in our mind that is constantly being influenced by those of others.

Most of our reality bubbles might prove to be very similar to those of other people - except for some details. The effectivity of communication depends on agreement about many things. Since our sensory input is mostly the same it makes it easier to agree on sensed phenomena.
To be useful an artificial reality model would have to resemble our understanding of the world. Meaning is added to information in order to supply a handle for intelligent processing.
Previously the question was asked if an intelligent system would need to be conscious of itself. Another step is to wonder if a system needs to pass a Turing test to be of practical use. The answer is probably no - but a large scale system as proposed should be able of passing the test and gradually improving the results.
Such a system should be able to learn. There are already systems available that can teach themselves based on feedback. Most exciting is the possibility that a single system can learn from all of us - that we can teach it as and raise it to full potential.

Building the foundation

When it comes to describing a system the first comes to mind are sets, subsets and intersections.

Tulips are a genus of plants belonging to the division of flowering plants. What this means is that objects are organized in groups according to properties. The subset adding specific properties and inheriting them from the ancestor set.

All plants use photosynthesis to create glucose. All flowering plants make flowers during certain reproductive phases.

Plants are objects made out of matter, just like rocks or masses of water. Objects of matter have a certain size, mass and a position in space and time.

Of course you could start describing a flower, or yourself and gradually arrive at the fundamental laws of nature. Or you could start with a ready made foundation based on classical mechanics and weave in all the details.

Classical mechanics describes the macroscopic world we all experience. Based on the concept of space as an arena for bodies to interact under the influence of a system of forces. . Since quantum mechanics applies on an atomic level or smaller, it's not as relevant to us or our reality system.

From clouds of gasses to bodies of water to solid objects. Thanks to our curious ancestors extensive mathemathical descriptions of how they behave under influence of any force is available.

Computer games often construct a universe within the game. Simulating Newton's third law (action = reaction) a ball can seem to bounce off a wall - the exit angle reversing the entrance angle.

Now imagine the same on a larger scale. A complete virtual universe as we know it. We're all part of it. A virtual you will have to be part of it too.


Economic validity

Cloud systems providing high processing performance would be in demand, as well as fast mobile networks.

Personal nodes in the network could share an amount of storage space and processing power with the shared system. Chunks of data can be redundantly stored in a distributed manner, much like peer-to-peer systems. That way, the more personal records you keep, the more you contribute to that shared systems.

Much of the needed software would have to remain in the public domain to prevent abuse.

More hardware and cloud processing will result in a more detailed virtual representation of an individual. As well as more services that can be provided. Controlling lots of robots for instance could require some extra processing. People who can afford it wouldn't be in need of humans serving them - and they might just appreciate the privacy or moral benefit.

A personal system should be assigned at birth to everyone - for society's sake - according to the earlier mentioned concept of assisted development and impersonal surveillance. A basic version could provide all of this, still leaving all kinds of optional features and extensions possible. Ideally the network of systems will grow because of it's usefulness - not because it's forced upon you. In the same way the internet itself is not controlled by a single government, a network of impartial systems is just a medium for all to use. The internet made it easy to lookup when the grocery store, or even order the groceries and have them delivered. A personal system would order them for you. Maybe even cook for you in an automated kitchen.


You might have to wait

We're talking about a hypothetical situation as long as electrical energy isn't abundant and inexpensive. Moore’s law still would have to apply: processing and memory capacities double every two years.

The shared reality bubble might prove to be a overly large-scaled system. Suppose every bit of information on the web is analyzed and fit into a model. There is a lot of redundancy out there of course. However storing information in a meaningful way would most likely require more storage space than the original information did.

A single personal system may require as much processing and data storage capabilities as currently handled by a whole datacenter now. Running all the other required systems - from speech analysis to routing all the access requests - will add up too. Suppose we need a 1000 times more storage and processing capacity, then following Moore's law we might have to wait another 20 years.