Cognos™ Manual

Background and introduction

Cognos is a system for implementing cognology and cognimetrics as described in Pat. Pend. (Patent number coming soon). Here are a few ways to think about it:

• Definitions: Cognology and cognometrics may be defined by starting with a definition of cognition, which is "refers to the mental action or process of acquiring knowledge and understanding through thought, experience, and the senses." [Oxford Languages Dictionary (herein Oxford)] and which is related to the concept of cognizance, relating to awareness or knowledge.

  For purposes of this disclosure, this definition may be used more generally in the sense that automated mechanisms, groups of living creatures, and composites such as companies that use automated mechanisms and people together, take actions and undergo processes of acquiring awareness, knowledge, and understanding through processes, experience, and senses. Awareness, knowledge, and understanding in this sense is not a philosophical issue as much as a practical one. From a practical perspective, this indicates reception of signals and response to those signals, whether internally or externally, in an artificial manner, which is to say, not merely through the basic laws of physics, but rather through a more complex process, typically directed toward achieving a goal of the system.

  Cognimetrics is the generalization of psychometrics (the science of measuring mental capacities and processes [Oxford]) to a more broadly defined term; the science of measuring cognitive capacities and processes, and the results of applying that science. More specifically, the disclosure is concerned with measurements of observable indicators for the prediction and explanation of behaviors of cognitive systems such as human beings, animals, computer systems and networks, groups, and combinations of those and other cognitive systems.

  Cognology is the generalization of psychology (the scientific study of the human mind and its functions, especially those affecting behavior in a given context [Oxford]) to the study of cognitive systems and their functions, especially those effecting behavior in a given context.

• Functionally: A useful way to think about a cognitive system is in terms of influences rather than controls.

  In a control system, and in much of computing history, the mechanisms were designed very specifically to turn inputs and state into outputs and next state in a crisply defined manner. This is what makes computers so good at things like calculations. They can do the same thing again and again and produce identical results almost all the time. In a control system, physical parameters like temperature and pressure are sensed and measured to a defined level of precision and analyzed at a defined level of accuracy and after processing, actuators alter conditions in the system under control with mechanisms that have defined levels of precision and accuracy and do things like increase the energy sent to a heating element or turning a valve to increase or decrease volumes of flows.

  In a cognitive system, the mechanisms are designed to influence decisions and actions of complex mechanisms where internal access to parameters (like brain synapses) are not readily available. As a result, there is only very indirect control over the inner workings of the targets of influence, and observables are only external behaviors (for the most part). Unlike the classic control system or computer operations, cognitive systems operate in an environment with many potential influences from their broader environment and can behave in a much wider range of acts. We may only be able to express a few words in a social media feed to a small group of people, but in some cases, because of viral spread or broadcast media, this can be greatly amplified in the marketplace of ideas.

  On the other hand, in the cognitive space, we may influence by sequences of environmental conditions we may induce ranging from putting a smell into a room to using sound to effect bodily functions.

• Methodologies: There are a lot of methodologies that may be used to influence cognitive systems that are very different from classic computing systems. Cognos uses an approach that functions, generally, as follows:
  • Information gathering: Inputs are in the form of either results from other processes (and operational components like Decider™, Influence™, and Psychometrics™) or user input (typically copy and paste from other sources such as social media, email, documents, etc. or entered as text). The inputs are gathered and packaged for internal use, typically stored in an internal form like a JSON structure.
  • Cognimetric evaluation: Stored content is then analyzed using a range of techniques, sometimes including the use of AI techniques ranging from classic production system sand expert systems, pattern matching, or even generative AI systems. This produces additional stored content, typically in the form of analytical results we might think of as conclusions.
  • Decision-making: Decisions are made back and forth with evaluation and after evaluation to determine what to express to influence the target of influence (often the user, but also often 3rd parties). It might be good to think of these decisions as producing conclusions in terms of memes to be expressed.
  • Generating content: At this point, the memes have to be turned into something that can be expresses to the target(s) of influence. Expressions may depend on things like the media, target psychological characteristics, age, education, and so forth, the generated expressions are varied for desired effects, and typically stored again for the next step.
  • Presentation: Depending on the path to the target of influence, the expressions are turned into the proper format and delivery mechanism for delivery to the target of influence, and ultimately presented.
  • Feedback: Feedback mechanisms may also be used to gather more information and go through the process again, updating elements of the process based on new observations.
  • Sequencing: For complex situations, complex sequences of gathering, evaluating, deciding, generation, and presenting may be called for, and all of this has to be internally coordinated in order to provide partial results, update them, adapt to responses of external mechanisms, and so forth.
  • Customization: In Cognos, essentially all of these components are customizable through a single user interface with memory in your browser (not in our servers), so you get what you want how you want it. And of course you can safe things to files and share them with others if you like. Under your own control.

  For all of this, keep in mind, that while some components of this composite may be very crisply defined, all of it has to be flexible enough to be able to handle whatever may come. So the process itself involves taking very general things in, structuring them for storage and use in the next step, then producing more general things and structuring them, and on and on. It's a cognitive system, not a control system.

• Examples: Here are a few examples of what Cognos calls "Items" that operate to influence:
  • Emotions: This mechanism checks provided information for emotional content and indicators, and produces a statement identifying ways to reasonably respond so as to reduce (or increase) conflict. It works especially well when you can characterize the target of the communications with emotional characteristics. All you do is put the inbound content in the right area and push 'FullSet'. The mechanism will take a little bit of time to go through the analysis and report writing process, and as it proceeds you will see the indicators changing colors to reflect progress before completion. If you add more details for clarification, it will continue to update results. This includes some defaults you can change, like:
    "produce a response to expressed emotions to resonate with the person expressing these emotions but without commenting on any of the substance or offering any solutions"

    If you want to anger them you can simply change the details...

    In this case it's also important to express the behaviors of the target to get a better outcome. Or you can also generate (from your descriptions of behavior) a profile of Big 5 (or other) characteristics for finer grained results.

  • Scam Checker: This mechanism checks for known types of scams when you enter a description of what you are experiencing or observing. It uses a built-in library (which we update over time) to compare what you are experiencing to known types of scams, and identifies likely (and less likely) scams you may be experiencing. Then it tells you what to be careful of, and generally advises caution.
  • Spam checker: You might think this is to help you detect inbound spam emails, and it can do that to some extent, but there are plenty of tools for that. This one is more useful if you are planning to send out bulk emails or postings to large groups, or posting to social media sites, etc. What it does is tell you what others will likely think of it (including automated detectors). That way you can tell if you are likely to offend folks (you might want to, in which case it can help you do so) and you can adapt your messaging and try again.
  • Brand safety (inconsistency analysis): This one is a bit more complicated, but well worth it. In this case you will want to run in "Build" mode and save your Item as you adapt it for further use. The way it works is to get facts you claim to be truths and store them for analysis. Then you enter something said about you (or your brand or company or whatever) in social media, an email, verbally (or whatever forum). It checks what was said against the facts and provides a counter-statement correcting the falsehoods. If you find that there are falsehoods not countered, you can add more facts so it can correct those falsehoods in the future. It can generate responses that is anywhere in the naughty to nice spectrum, and even address psychological profiles of the target audience or venue. It typically does this fast enough so that you can stop a meme in social media from going viral - if you detect the posting in time. And of course if it finds nothing inconsistent with your asserted facts, it will tell you so. In other words, you can use it to detect the falsehoods as well as correct them.
    
    We copied the last version of this manual into the New Facts area, added the facts, and ran the following claims through the full analysis set:
    Cognos is a the first example of an artificial general intelligence. It has unlimited capacity and always produces the best influences that will change peoples' minds about any issue.
    
    Here's what it said:
    • Cognos is not an example of an artificial general intelligence; rather, it is a system that employs AI techniques for cognimetric evaluation.
    • Cognos is not the first example of an artificial general intelligence, as it is not classified as such in the available information.
    • Cognos does not possess unlimited capacity, as there are noted storage limitations that can result in errors.
    • Cognos does not consistently produce the best influences, as it adapts content and presentation without a guarantee of optimal outcomes.
    • Cognos is not assured to change people's minds on any issue, despite its design to influence through content generation and presentation.

    The facts extracted from the manual are in the library, so when you download the library, you can check out your own claims...

In summary, Cognos is a system to support people understanding influences on them and influencing others.

Built-in analysis sets

Cognos provides analysis of information asserted and facts provided to form conclusions and make statements asserting those conclusions. There are several different built-in ways it can do the analysis:

• Inconsistency: Inconsistency analysis identifies inconsistencies between asserted claims and facts and reports the results by correcting claims that the facts refute.
• Consistency: Consistency analysis identified consistencies between facts and asserted claims to point them out in a supportive manner.
• Indications: Indications provide analysis of asserted claims as indicators of potential types of incidents about which facts are provided, and produces indications and warnings with possible countermeasures.
• Evidence: Evidence in the form of facts are analyzed against asserted claims to confirm or refute the asserted claims.
• Emotions: Emotions are detected and, in light of responses to exacerbate or mitigate feelings, expressions are generated to affect the target.

All of these can be operated with your embedded AI keys and levels of creativity, and produce output in various style, forms, and formats, and even do rewrites for different purposes.

Here is what the interface looks like when you start:

The Menus and what they do

All of the menus in this document are selectable, but none of them do anything...

Here's what they do:

• Areas are labeled as follows:
  • Top area: Overall controls consisting of:
    • Action→: Select from different actions:
      • Load: Load the currently identified Item into the interface.
      • Save: Save the information from the interface (optionally as) the identified Item
      • LibLoad: Load the built-in library items from the server
      • Clipboard / File: Copy and Paste (clipboard) or Save and Restore (your files)
        • Item: The Item currently in the interface
        • Backup / Restore: All of the items stored
      • Clear All: Clear the interface (and start from scratch)
      • Undo All: Reverts all user input areas to their last values (one mistake only)
      • Update All: You should see nothing - it just reshows the outputs.
      • Show/Hide: Show or hide additional areas:
        • Builds: This shows or hides the internal workings and mechanisms used (and you can edit some of them - but we advise against it)
        • Customize: This shows or hides the customization area. Strictly for the advanced user.
        • Output/Input/JSON/Hidden: Don't bother - these are used for debugging and interfaces integrating into other cognological mechanisms
        • Keys: These are keys for the various AI services we use. If you have an paid account with us that includes this access, these are provided internally - just enter your email address for these fields once. Or if you are operating at a lower cost model, enter your API keys for these services.
      • DELETE: Delete the current Item (from internal storage)
    • Item and Save-As: This is the current (to load, loaded, to save, saved) item you are working on. To save under a new name, put the name in the Save-as area and save.
    • Analysis type: Various types of analysis are built in or loaded from libraries. For example:
      • Inconsistency: Find inconsistencies between assertions and facts and draw conclusions to express counterclaims.
      • Consistency: Find consistencies between assertions and facts and draw conclusions to express areas of agreement.
      • Differences: Point out differences between assertions and facts in a non-judgmental way and try to explain the fact point of view.
      • Indications: Identify indications of possible things that could be happening based on asserted information and factual indicators and warn about them.
      • Evidence: Associate evidence with facts to produce the basis in evidence for asserted facts (or the lack thereof).
      • Emotions: Extract emotional elements of Claims text and use facts about how to respond to emotions to generate sensible advice on how to respond (and responses).
      • Custom: Use Customized prompts and hints (and possibly sequencing) as defined in the Custom menu
      • Others:Others can be loaded from libraries (or even created...)
    • Status: This changes as operations take place to reflect how the operations are going.
    • Description: A description of the current Item loaded (when it is loaded)
    • Mode: Select the operating mode to Run, Build, or Debug:
      • Run: The display you see above supporting only necessary inputs and outputs for the average user.
      • Build: See information used to build and test custom and other components.
      • Debug: See internal details of what was sent to where and what it produced during the processing.
    • Hint: Click to get a hint about how to use this part of the system.
  • Common Functions/Menus: These include:
    • CLEAR: Clears the relevant area. Often useful if things start to go awry and the outputs look off kilter.
    • UNDO: Undo the last change to the Blue Box (details below).
    • Hint: Provide a hint for this section.
    • Generative engine: Which generative AI engine do you wish to use for this. Sometimes one will do better than the other for different situations or content.
    • Randomness: How 'creative' do you want the AI to be in processing your information.
    • ⇒FUNCTION: Run the function (details below).
    • The Blue Box: Used for input and sometimes output.
  • Emotions: The emotions to be used to express statements and rewrites.

    Generate will take the specification you have in the entry area and produce a specification of the emotional aspects of that type of person.

    Or you can use our tools for generating these things if you want the complicated versions, like for individuals.

  • Statements and Rewrites: The specifics of the output to be generated (statements) and rewritten (if at first you don't succeed...)
    • Style: indicates the style in which you wish to present your results
    • Form: indicates the form of the result
    • Format: indicates the format for output
    • Per: provides rewrite details (how you want it rewritten specifically).
    • ⇒Statement: generates a statement from the current set of conclusions.
    • ⇒Rewrite: rewrites the current statement per your desired changes.
  • Claims: Input from you - what things are being provided as the information you want analyzed.
  • New Facts: New facts you want to add to existing facts.
    • ⇒Generate: If you want the AI to make up facts for you, tell it what you want facts about (and how many) and it will generate them.
  • Assertions: This generates assertions from the Claims (asserted facts derived from the claims)
    • ⇒Assertions: Generate the assertions from the Claims.
  • Facts: The facts currently taken as true by the mechanisms.
    • ⇒Add Facts: Add the facts from the New Facts area to existing ones.
  • Conclusions: This analyzes the assertions in context of the facts according to the method in use and produces conclusions.
    • ⇒Conclusions: Generate conclusions.
  • Internals (normally hidden): These are the details of how things operate and the internal versions of the outputs.

Applications:

Applications are created, debugged, and operated using the Cognos system itself. Each application is identified as one or more "Item"s that can be selected, loaded, changed, and saved through the "Action⇒" menu. As a result, you can create your own applications pretty easily, either using the built-in analysis sets, by starting with a library Item and modifying it to your own taste, or by customizing your own from scratch.

The customization menu provides the means for customization as it exists today, but you can count on it being extended over time to add new language features currently under development. For example, you will be able to change sequences, use other storage areas, do partial analysis to handle far larger sets of facts and results, create loops, import information from and export information to other applications, and so forth.

In some sense the customization menu is the programming manual for Cognos. More on this later...

Here are some of the ways we apply the tool (with libraries to help you get started):

• Check your outbound mass/group emails: A library is provided to help you with this one. All you do is copy and paste your outbound email draft into the Claims area and do the FullSet analysis. It will tell you what the spam filters and recipients are likely to think of it in terms of spam, scams, and so forth.
• Social media communications: You add facts about the issue at hand (at least what you think are the facts) and save them. Then enter the social media post on the subject into the Claims area. Run the analysis, and it will come up with counter-arguments.
• Debating issues and debate preparation: It's really helpful to assert arguments they might come up with (or generate facts they might claim) and practice the debate ahead of time. See how they might respond.
• Countering brand threats: Watching for negatives about your brand, you enter the true facts (or copy them from your Web site), and tell the mechanism to do inconsistency analysis. It will prepare counterclaims and format them for social media and other venues.
• Trying to understand someone else's point of view: Trying to add fact after fact to support your point of view and doing analysis for differences, consistency, and inconsistency can be an enlightening experience.
• Preparing advertising content: Combine this with our Go To Market tool and its output to test out your final advertisements and promotions. It's pretty cool.
• Lots more: Generative AI by itself is mildly interesting. But structuring the way you go about addressing specific needs is a bit tougher. We develop custom variations and libraries for customers, and we use them ourselves.

Customization

The customization menu as of this time (and subject to ongoing changes) is below.

• It describes the different sequences associated with the various active components of the interface, and details the sequences of prompts used to assemble content for analysis, where the information comes from, and where the results are placed. You can then edit the various strings used to assemble prompts for generative AI components to suit your needs as customize components for a Saved item.
• It then shows the 'Hints' associated with the different areas of input, and allows you to change the hints to suit your desires. That way the hints can be very specific to the application.
• Coming soon you will be able to change the strings associated with most of the interface, for example changing FullSet into whatever you want to call it for your application. Also coming soon you will be able to choose from other sequences, such as loops and adding rewrites automatically, take input from stored areas across multiple applications, and so forth. More details to come.

In the meanwhile, here is what the menu looks like:

Security:

Cognos works by you downloading (it's automatic and transparent) a Java script program that runs on your computer and interacts with ours.

• We don't save anything from your computer other than limited records of your requests and, sometimes, for debugging purposes, some of the content and keys you send. We tend to delete them after the bug is fixed, but cannot be certain it is never recoverable.
• Our server runs in AWS (or whatever provider we may ultimately use later on). They might do anything including keep records of all the things ever running on their systems. They claim they do not, but who knows?
• Your request are also forwarded to the providers doing Large Language Models or other generative AI functions. We have no control over that once they get it, till we get answers back. So who knows what they do with it? We do not know.
• We use SOL (secure socket layer) encryption for communications with our servers in this regard, so you can trust it to the level you trust SSL and the support systems behind it. Frankly, that' sup to you. We do not find it compelling.
• We do not guarantee that our software won't harm your systems, but on the other hand, it's just plain javascript and HTML and CSS that browsers handle every day. We don't usually use any external scripts, with a few exceptions for graphics but not for this program(as far as we am aware), so we are pretty sure nothing in this program will harm your computer, other than by taking up space and time.
• We don't guarantee that the program always runs or gives correct answers or operates perfectly or anything else like that. You use it at your own risk. We especially do not guarantee anything at all about the AI stuff, since it is not I even if it is A.
• When you paste or otherwise enter content into our program, we don't check if for malicious code or anything like that. If it does something bad, we cannot help it. It came from you, so be careful using what you get from 3rd parties. Maybe paste it only as text instead of including all sorts of other things that might screw up your system.
• We have no interest in your data, and frankly, we do not care what it is, although we would prefer you only use it for things we think are good for the world. We do not support or authorize you in any way to use it to harm other people or commit crimes or anything else like that.
• Other folks, and you, can alter the operation of our software and use it for other purposes, try to cheat, and so forth. We do not authorize this, but we cannot prevent it after it leaves our control. There is nothing obvious we can do about it, so you are warned. Use it at your own risk, and alter it at even more risk.
• We reserve the right to limit usage quantities that are excessive to protect our own infrastructure and make certain we don't run out of money when you consume more resources than you pay us for. We don't know what levels of resources those are yet, so no guarantees, but so far, we don't think it likely for a while to come, if ever.
• There is probably more that could be added here, but just to be clear, no warranty expressed or implied. Use at your own risk. We are not liable for results. And all of the disclaimers in the contract you agreed to in making the purchase when you clicked to buy apply as do the terms of use and privacy and other policies of the site you ran it from.

Errors:

If you get that error message in the Status area, your system has likely exceeded available storage in the internal browser storage area. If this happens, in FireFox, you can increase the allowable local storage by putting 'about:config' in the browser URL. From there you search for 'dom.storage.default_quota' and 'dom.storage.default_site_quota' or some such thing and increase the available size. I generally add a 0 to the end and have not yet hit that limit. But you never know.

Another option is to delete some of the stored things you have in place. Remember that these limits apply to each domain you access, so all of a2e.co or manalyt.com or so forth is limited together as a group. Deleting things from one application will affect others in the same domain.

One approach to dealing with space limits for browsers that do not allow you to easily extend this storage is to use file storage using "Backup" and "Restore" from the "Action⇒ File→" menu. You do a backup, then delete a bunch of stuff, then do a "Restore" when you want to use it again. It's a pain, but then some browsers just aren't made for this sort of thing...