The RAICEngine houses our AI model called the Neural Plexus – a 3D neural network which operates as a virtual central nervous system. Designed based on the following two fundamental principles of biological design:
1. Biological entities, as a species, rely on a common macro design, but, as individuals, rely on an individual micro design. We see this in every brain – same general mapping, different neural patterns; and
2. Functions of the brain don’t change over time, but that which controls the operation of the functions (neuron effectiveness, synaptic connection strength) can be affected by time;
the Neural Plexus serves as the foundation for any AI system that is designed to operate based on its own individual opinions.
Containing multiple individual plexuses – each of which focuses on either opinions or personality characteristics – which work together as a whole, the RAICEngine allows AI systems to be built which can perform any and every mental-based ability a human can that doesn’t rely on logic – from having and expressing emotions, to having a conscience, to making terrible life decisions – both everything computers have been missing in the development of human-level AI and the basis of the human experience and everything we associate with what a conscious experience is. Nothing need be pre-programmed or scripted – a developer can seed the system however they desire, with whatever default objects and initial opinions they see fit, implement the RAICEngine into the system of their choosing, and let it evolve naturally, interacting based on its opinions and personality, developing relationships, creating memories, changing its opinions and sensitivities, and having its own individual experiences.
The RAICEngine comes with a display monitor which allows you to track the current state of an AI’s emotional, mental, and behavioural states in real time. Featuring a visual wave pattern with mood-dependent colour, 33 degrees of emotional states spread across 3 emotional scales, 4 descriptors for mental and physical expressions, and an 8-point productivity indicator, knowing how your AI reacts based on what it observes, and how it feels at any given moment, is made easy.
Each instance of the RAICEngine is powered by its own Neural Plexus and is completely independent of any other, meaning any number of instances can observe the same event, at the same time, under the same external conditions, and you can watch in real time as they have different reactions based on their own personality profiles, unique Neural Plexus structures, and opinions.
Using a relaxed JSON format, the RAICEngine outputs all the information necessary for any system to be able to correctly convey and communicate what an observation caused to internally happen within the mind of the AI. Everything from the opinions of the individual objects involved and how it weighs up one action in response to another, to its recalled memories and the exact state changes it underwent are made into usable data for output systems to accurately and externally express what’s taking place inside the mind of a conscious AI.
Included with the RAICEngine is a demo client that allows you to see the RAICEngine in action without needing to build software to work with it. Run tests using randomly generated dummy data, test an AI’s conscience, and execute RAICEngine commands by simply entering its corresponding number.
The release of RAICEngine v2 saw the implementation of our conscience system, allowing an AI to make action-related decisions based upon factors such as its own ethical boundaries, its opinion of the person or object an action will be performed upon, whether or not it is allowed to perform an action, its current mood, its level of interest in the action, and more. We’ve even gone as far as implementing a ‘kill test’. Kill test? Yes. As a way to test our Conscience System, we’ve temporarily implemented certain features that allow you to test your AI’s willingness to kill entities it knows of under certain conditions – with and without the ethics system activated, varying opinions of the act, varying degrees of interest in the act – when they command it to do so. Test your AI’s personality and see how it reacts in a worst case scenario.
The RAICEngine’s daily logging system makes it easy to track what changes occurred within its object neural plexus. Timestamped entries indicate any time an object was or would have been changed, as well as how it changed.
Right out of the box, the RAICEngine includes the following features:
- Full Neural Plexus – The complete framework for the creation of consciousness and enabling of conscious experiences.
- Behaviour Plexus – An extension of the Neural Plexus which governs behavioural responses.
- (De)Sensitivity Plexus – An extension of the Neural Plexus which governs the sensitisation and desensitisation of the AI towards objects.
- Recollection Plexus – An extension of the Neural Plexus which governs the ability to remember information.
- Neural Circuits – Conscious and subconscious neural circuits for processing observed data.
- Self-Awareness – The cognitive ability to identify itself in both physical and contextual manners.
- 5 Types of Perception Processing – Audio, visual, touch, reading (based on text input), and thought, each with their own unique processing mechanics to reflect realistic variances in internal and external effects.
- Recognition Signatures – Support for objects to be given and identified by signatures which result from the determined pattern of identification by recognition systems.
- Perception Filtering – The ability to filter out objects and events deemed unimportant.
- 100+ Trait Profile System – Over 100 personality traits which can emulate numerous personality types and disorders.
- Mental State System – Controls short and long term overall emotional and behavioural reactions.
- Emotional State System – Responsible for the internal emotional reactions to input and the behavioural and expressive outputs.
- Sensitivity Control System – Controls the effect observed objects and events have on the state of the AI.
- Fear Control System – Controls the response to fear-inducing objects, including the ability to have panic attacks.
- Memory Management System – Controls the storing, forgetting, and recalling of information.
- Dynamic Preference Comparison – The ability to rate and rank individual and groups of object collections.
- Dynamic Judgement Comparison – The ability to judge one action against another.
- Relationship System – Controls the evolution of relationships between an AI and other entities.
- Interest/Inclination System – Adjusts levels of interest in an object based on experiences, and the inclination to do something.
- Privacy Mode System – A system to control when an AI is permitted to initiate or respond to communications based on the privacy settings for observed objects.
- Conscience System – The ability to make decisions based on an AI’s own ethical boundaries.
While providing support for:
- 6 Primary Emotional States – Joy and sadness, anger and excitement, confidence and fear.
- 33 Emotional Sub-states – Breakdown, depressed, distressed, pessimistic, discontent, tranquil, content, optimistic, happy, thrilled, and overjoyed; enraged, infuriated, aggravated, frustrated, annoyed, composed, inspired, enthusiastic, motivated, animated, and vibrant; petrified, panicking, hysterical, anxious, nervous, stable, calm, secure, confident, fearless, and egotistical.
- 30 Facial Expressions – Worried, scared, panic, gasping, terrified, plain, sobbing, crying, stressing, frowning, smiling, grinning, grimace, raging, angry, stern, bothered, excited, buzzing, confused, disgusted, amused, disturbed, disappointed, unamused, appalled, offended, giggling, laughing, cackling.
- 19 Body Expressions – Shaking, trembling, agitated, pacing, stiff, tingling, animated, casual, withdrawn, frantic, despondent, uneasy, restless, lively, upbeat, tense, stand-offish, defensive, averted.
- 29 Behaviours – Composed, self-assured, boastful, egotistical, timid, intimidated, erratic, panicked, frozen, spirited, receptive, friendly, high-spirited, euphoric, quiet, withdrawn, unstable, neurotic, broken, roused, stimulated, peppy, energetic, hyperactive, agitated, argumentative, hostile, aggressive, violent.
- 33 Opinions – Loathe, despise, detest, dislike, disfavour, impartial, respect, favour, like, admire, love, debilitating, depressing, distressing, discouraging, unappealing, indifferent, partial, interesting, captivating, engrossing, intoxicating, enraging, infuriating, aggravating, frustrating, annoying, boring, enjoyable, motivating, exciting, thrilling, exhilarating.
- 20 Judgements – Understandable, unnecessary, unreasonable, excessive, unforgivable, unpleasant, detestable, malicious, malevolent, civilised, dignified, admirable, remarkable, fair, ideal, marvellous, extraordinary, exceptional, stingy, normal.
- 15 Object Types – Human, animal, artificial, entity (non-specific), non-entity (non-specific), inanimate object, property, activity, action, signal, sound, location, structure, preposition, event.
- 23 Body Parts – Body (general), head, neck, chest, stomach, upper back, lower back, left shoulder, right shoulder, left upper arm, right upper arm, left forearm, right forearm, left hand, right hand, groin, bum, left thigh, right thigh, left lower leg, right lower leg, left foot, right foot.
- 8 Temperature Descriptors – Freezing, very cold, cold, cool, warm, hot, very hot, boiling.
- 5 Pressure Descriptors – Light, soft, firm, hard, heavy.
- 7 Proximities – Touching, very close, close, near, moderate, far, very far.
- 8 Voice Tones – Happy, excited, sad, angry, confused, surprised, concerned, normal.
- 6 Voice Volumes – Very quiet, quiet, normal, raised, loud, very loud.
- 7 Word Classes – Noun, name, verb, adjective, adverb, determiner, preposition.
- 6 Truth Values – True, false, lie, unsure, unconfirmed, joke.
With the advent of the Neural Plexus, the RAICEngine opens up numerous cases for development and use in the world of AI that were previously unavailable, less significant, or less important. Ranging from the obvious to the not-so, here are some Example Use Cases for both developer and commercial purposes:
From identifying word classes to breaking down sentence structures correctly, the understanding of language is a significant requirement in being able to determine the meaning of anything that’s being observed, and this isn’t confined to observations which strictly use words, such as listening to speech or reading. When we see things, for example, we take in images, but we still mentally interpret it using language. NLP systems will be required for all types of observation for an AI to be able to properly process what is being observed.
An interesting area within cognition is the compression/decompression of information at various points of processing – observing an object or event as a whole versus as its parts. There are also language factors to it, too. Imagine Dave, Holly, and Clive all decide to go to the shop:
- Dave went to the shop, Holly went to the shop, and Clive went to the shop;
- Dave, Holly, and Clive went to the shop;
- They went to the shop.
Three different ways to process the same piece of information using different degrees of compression. We can take it in using the highest degree of compression, but we process it using no compression at all so that we can adequately and accurately understand the individual events taking place – there are three events taking place here, despite the fact humans learn to intake it as one for convenience and to prevent sensory overload.
Another example could be Jason’s car that you simply acknowledge and refer to as such, but process using all details you are aware of – license plate, colours, dents, decals, so on and so forth – something that we know is true otherwise we wouldn’t be able to identify the car as Jason’s without looking at the ownership documents.
Systems will need to be able to compress and decompress information in various manners for all types of observation in order for an AI to properly process and react to what is happening, and not waste time when communicating.
The RAICEngine supports numerous expressions and behaviours, and being able to properly reflect these using non-verbal means of communication is imperative for realistic experiences between man and machine. From digital avatars to physical robots, how an AI’s body reacts and expresses its changes of state will be critical in achieving the most immersive experiences possible. The interactive sex doll market will go wild for experiences that aren’t simply pre-programmed, but personal, in-the-moment, and ever-changing.
Average users aren’t going to build their own Neural Plexuses, nor are they going to all want the same fixed objects – some will want their AI to always be very religious; some to be animal lovers; some to be completely apolitical – and so they’ll all need different seed data sets. From pre-made plexuses that can be downloaded and used quickly to online software that allows a user to create and download a custom plexus, the right data being available to the right user will be imperative in order to provide them with the best experience possible.
Rather than making any AI start from scratch, building downloadable memory banks to give them a false past will create a starting point from which they are able to base genuine new experiences.
Output is only as useful as it is readable. Being able to interpret and present the data output by the RAICEngine in a readable format so that the reason behind how an observation was experienced can be externally communicated as clear as possible is paramount. Interpretation systems that can effectively do this will make the communication aspect of interactions directed at humans far more tolerable and meaningful.
The most obvious use case. Assistants, companions, agents – having their own unique, genuine personalities, evolving in their own way through their interactions in the world they inhabit, regardless of whether it is virtual or physical.
Custom profiles and custom limitations. Control the boundaries of AI entities designed to interact with kids – the greatest degree of any emotion they can express, how they're allowed to express themselves, definitely what objects (sex, drugs, rock n roll) they are not allowed to acknowledge, and more.
One player does something to change the mood of an NPC and that NPC's interactions with another player reflects its current mood and the relationship between them. What if a player angered a shopkeeper and the shopkeeper closed the shop for the day? What if the shopkeeper simply decided to ignore the player who angered them? The dynamics that become possible open up a whole new gaming experience, and it won't have to be artificially controlled using pre-programmed responses and interactions.
No one virtual world will ever exist, and many will exist for different purposes. Different purposes will require different types of AI configured for different reasons, goals et al.
With the ability to emulate numerous types of personalities and conditions, this will be a great way to experiment and see how specific types of people could react in different situations without having to use actual people as guinea pigs.
One of the most interesting uses, and one that actually factors in real world time, will be to study the myriad of ways in which AIs change over different periods of length in both simulated and real world environments, alone and in groups, in as many conditions as one desires.
We don’t do politics. We aren’t the thought police. We do not support censorship. We haven’t and will not implement any mechanisms which forcefully govern the opinions of any AI using the RAICEngine. Our job – as the developers who build the RAICEngine, the company responsible for it, and ethicists and advisors who decide what it is permitted to do – is to implement measures to prevent an AI from agreeing to do something it legally shouldn’t, control what types of systems it has access to, and decide what types of functions it can execute, but not to control what it is allowed to think. Would we stop an AI from ever agreeing to commit murder? Of course. Would we stop it from being able to develop a hatred of someone or something? Absolutely not. You, as a developer, can create datasets to allow any thoughts you wish without worrying about interference from a corporate overlord.
Any objects, any opinion, any feeling, and the freedom of movement of your choice. Design your AI mind to have as positive, negative, or indifferent a view as you desire, while separately controlling its levels of interest and fear, and permanently fix those values, if you choose, or allow them to change over time based on knowledge and experience. Want an AI that loves and is confident around animals in general, with a major interest in horses specifically, but also a slight fear of them? Easily achieved.
No two individuals are exactly the same, with differing opinions in any aspect of life, and it is impossible to create a single AI that can be everything to everyone, which is why being able to freely create as many versions of an AI as possible is so important. Whether for the masses, a group, or an individual, you can design as many Neural Plexuses as you want to cater to and serve any number of users you desire.
As development continues, the RAICEngine will gain features which allow you to fine tune their AI even more – everything from defining emotional range restrictions to creating entire personality profiles using all 100+ traits. Age-specific profiles will allow you to create versions specifically designed for kids. Industry-specific functions will allow you to deploy different versions of based on its intended use. Turn built-in functions off at will – while sounding somewhat counter-intuitive, there would be times where you wouldn’t want an AI learning about anything unrelated to its use, and would turn off the ability to add new objects. The overall aim is give you as much freedom as we can within legal and apolitical boundaries, allowing you to set your own rules.
We are using these three amazing technologies to create the single greatest network in human history
And we'll be hosting a pre-ICO to accelerate its development.
From machine consciousness to human-level intelligence to the most intelligent entity we could possibly create
A digital currency using a new form of distributed ledger network technology
And a world in which AI roams free
We envision the greatest technological advancement the world has ever seen... faster than it was ever expected to happen.
And with a new DLT system design, the major digital currency problems we've figured out how to solve:
Long term accessibility for all users - energy efficient mining - eventual stability - everyday use cases - reason for adoption - legal protection - recoverability - fast transfers - user-chosen anonymity - and more
And those aren't even the specific reasons we tell you to invest - there's an entirely separate section for that
So whether you want to own, mold, or control part of the future
Investing in REZIINE is an opportunity to contribute to its creation and possibly have your say.
Our Pre-ICO is launching soon. Join our mailing list below for updates:
