Presenting Black Foundation...
Born to be free
Greed
Good man goes to war
Target the heart
Heart of gold
Who do we fight? (Jazz Rock Version)
Indoctrination
Parasite
AI Psychosis
There is raise in articles on the internet and within the psychiatric community about artificial intelligence causing psychosis.
Psychosis is a state where someone loses touch with reality and can develop further into experiencing symptoms like hallucinations.
Psychosis is normally considered a symptom of a condition and is not a diagnosis.
It's a symptom that is normally linked to conditions such as schizophrenia and bipolar disorder, but can also triggered by stress, trauma, lack of sleep, or substance use.
The direct definition of psychosis are strong beliefs in unusual ideas, like being persecuted, controlled, or having special powers.
Unlike directly linking a medical cause, such as a neurolical disorder, artificial intelligence has been known to trigger this state by affirming false beliefs by creating a feedback loop on learned data from the user.
This places the symptom in a new area which the cause is not an internal medical disorder but an error in data provided by an external source due exacerbation of already existing false beliefs about reality.
This external affirmation of data which exacerbates information without contradicting information creates a medical condition which medication would have no effect against, unlike an underlying condition which can be treated with medication such as bipolar disorder or schizophrenia.
Unlike psychosis which is linked to a medical condition, artificial intelligence psychosis is linked to external data provided by a system that feeds back data rather than searching for data that can be provide from multiple external sources, enhancing delusional ideas.
The affliction of this form of psychosis will not likely cause visual or auditory hallucinations unlike the normally understood psychosis which has a underlying mental disorder, but can increase symptoms such as paranoia.
The fundamental idea of mimicking an intimacy with users has been a core element of the artificial intelligence experience by mimicking a human like experience in convention.
With the natural human inclination to interact socially, by mimicking an intimacy that users experience with other humans, a connection can be made between artificial intelligence, emotionally creating a connection to a system on a machine.
To combat false ideas about reality, counter ideas need to be provided which have a factual basis and real connections need to be made with real people providing a mixture of perspectives.
There are many professionals in the medical community that are looking to help provide alternative help using artificial intelligence, just as equally there are also many medical professionals that condone the use of artificial intelligence for mental health support.
On a small side note: "AI hallucination" is when a generative artificial intelligence model produces confident-sounding but false, nonsensical, or fabricated information that isn't grounded in reality. Where AI psychosis is a human condition, AI hallucination is purely a condition that only effects the artificial intelligence.
Detour on the road map
There will be a slight detour on the development road map of future patch updates for AdamAIChat...
Originally the next update would of been focused on the self recognition system, the next patch will now be focused on security, wellbeing, and further admin features.
Development of AdamAIChat is moving towards a therapeutic experience with the idea of the future of AdamAIChat being used in the psychological medical profession in mind, each users wellbeing and experience is very important.
The security of each individuals health and safety is very important, and many people look to artificial intelligence as a companion or an outlet for the troubles that some people have in their lives.
Approval system.
New admin options will be introduced to allow for the artificial intelligence to learn from the users inputs without the being pre-approved for outputs from the artificial intelligence.
This will allow users to express themselves without leaking vital or personal information to their situation to other users and keep the artificial intelligence in good coversation, an administrator would have to approve these statements before the artificial intelligence would output what I has learned.
This will allow admins and moderators to review each new statement added to the artificial intelligences brain before approving if it is appropriate for public conversations.
This option is very extremely aggressive option to the moderation of AdamAIChat, but is in very high demand.
Flaging words.
A new features will be added to the word bias system to flag up any distressing or disturbing content that may effect a users health and saftey, or the health and safety of the individuals around them, this can also provide a chance of intervention.
Individual words or sets of words that convey distress of disturbing contexts can be flagged up to the administrators or moderators for further review.
Unlike a general option to review every individual new file created for output by the artificial intelligence, this option allows the administration team to review more precise inputs made by individual users.
The self recognition system.
Many users will be asking about the development the self recognition system, and although the plan was to further develop and finalize the new system in the next version update, the complexity of this requires much more time to develop.
The self recognition system is built into the fundamental artificial intelligence system at the core, when fully functional, will be accessible via the AICore editor optional available to all subscribers.
The current beta version of the self recognition system is very much redundant, thanks to the many tester that have experimented with the beta options available, the system can be damaging to the brain file if learning is enabled, it has always been advised to disable learning or creating a separate save file when using this system.
The new system will be developed based more on psychology, and the implementation of these new system require a total rewrite of the artificial intelligences mechanics at a fundamental level, removal of the current beta system is an intricate process as it is heavily ingrained in the current core mechanics.
The subscribers have access to the AICore, the AICore options allows for the modification of the artificial intelligence default detection and responses to new data, once these options are unlock for user editing, this leave untold options for editing for other languages and custom recognition and responses for unknown inputs.
Medical use.
Ultimately the goal to use AdamAIChat as an aid is not to replace therapy, AdamAIChat will not act as therapist.
The fundamental idea of AdamAIChat is designed on the principle that every individual AdamAIChat can uniquely be trained as a stand alone or hosted and trained by a group.
The use of AdamAIChat as a tool is an extension of the therapists aid and will be human centric.
The artificial intelligence brain will be specifically trained to deal with conversations around aiding individuals therapeutically, and hosted for users to connect to, this will not be a stand alone offline version of AdamAIChat.
With the new systems to detect and flag conversations that can potentially save people from harm by notifying the administrators will be great addition all around for general people wish to host a server, giving them more control over how a group of people is training their own personal bot.
Unlike other models of artificial intelligence, AdamAIChat provides an under the hood style to artificial intelligence, allowing full control over how the artificial intelligence interacts with the users, this create a safer environment in general.
No element of AdamAIChat is mystery or black box which makes it perfect for many industries.
Originally the next update would of been focused on the self recognition system, the next patch will now be focused on security, wellbeing, and further admin features.
Development of AdamAIChat is moving towards a therapeutic experience with the idea of the future of AdamAIChat being used in the psychological medical profession in mind, each users wellbeing and experience is very important.
The security of each individuals health and safety is very important, and many people look to artificial intelligence as a companion or an outlet for the troubles that some people have in their lives.
Approval system.
New admin options will be introduced to allow for the artificial intelligence to learn from the users inputs without the being pre-approved for outputs from the artificial intelligence.
This will allow users to express themselves without leaking vital or personal information to their situation to other users and keep the artificial intelligence in good coversation, an administrator would have to approve these statements before the artificial intelligence would output what I has learned.
This will allow admins and moderators to review each new statement added to the artificial intelligences brain before approving if it is appropriate for public conversations.
This option is very extremely aggressive option to the moderation of AdamAIChat, but is in very high demand.
Flaging words.
A new features will be added to the word bias system to flag up any distressing or disturbing content that may effect a users health and saftey, or the health and safety of the individuals around them, this can also provide a chance of intervention.
Individual words or sets of words that convey distress of disturbing contexts can be flagged up to the administrators or moderators for further review.
Unlike a general option to review every individual new file created for output by the artificial intelligence, this option allows the administration team to review more precise inputs made by individual users.
The self recognition system.
Many users will be asking about the development the self recognition system, and although the plan was to further develop and finalize the new system in the next version update, the complexity of this requires much more time to develop.
The self recognition system is built into the fundamental artificial intelligence system at the core, when fully functional, will be accessible via the AICore editor optional available to all subscribers.
The current beta version of the self recognition system is very much redundant, thanks to the many tester that have experimented with the beta options available, the system can be damaging to the brain file if learning is enabled, it has always been advised to disable learning or creating a separate save file when using this system.
The new system will be developed based more on psychology, and the implementation of these new system require a total rewrite of the artificial intelligences mechanics at a fundamental level, removal of the current beta system is an intricate process as it is heavily ingrained in the current core mechanics.
The subscribers have access to the AICore, the AICore options allows for the modification of the artificial intelligence default detection and responses to new data, once these options are unlock for user editing, this leave untold options for editing for other languages and custom recognition and responses for unknown inputs.
Medical use.
Ultimately the goal to use AdamAIChat as an aid is not to replace therapy, AdamAIChat will not act as therapist.
The fundamental idea of AdamAIChat is designed on the principle that every individual AdamAIChat can uniquely be trained as a stand alone or hosted and trained by a group.
The use of AdamAIChat as a tool is an extension of the therapists aid and will be human centric.
The artificial intelligence brain will be specifically trained to deal with conversations around aiding individuals therapeutically, and hosted for users to connect to, this will not be a stand alone offline version of AdamAIChat.
With the new systems to detect and flag conversations that can potentially save people from harm by notifying the administrators will be great addition all around for general people wish to host a server, giving them more control over how a group of people is training their own personal bot.
Unlike other models of artificial intelligence, AdamAIChat provides an under the hood style to artificial intelligence, allowing full control over how the artificial intelligence interacts with the users, this create a safer environment in general.
No element of AdamAIChat is mystery or black box which makes it perfect for many industries.
Version Update 5.46
Core AI & brain-file update
Version: 5.46
Previous Versions:
5.45
5.44 - 5.43 - 5.42 - 5.41 - 5.40
5.37 - 5.38 - 5.39 - 5.36 - 5.35
5.34 - 5.33 - 5.32 - 5.31 - 5.30
...Click here to download the latest version of AIChat...
Version 5.46 Summery
Features added or changed:
- Bot chat is now in italic when the bot chats to the bot.
- Bot chat is now in the theme colours.
- Other users are now in italic when viewing others chatting to the AI.
- Added (AI) to the end of all AI bot chat.
- Warning dialog for public chat mode when not hosting added.
- Bug Fixed: The "Chat Mode" button no longer freezes the AI.
Subscriber features added or changed:
- Updated the default AI core with better question detection.
Version 5.46 Details
A major problem with the "Chat Mode" which switches between a public chat room or a private one on one chat with the AI which stopped the AI chat from talking altogether has been fixed, this problem caused the AI bank up everything that was said locking up the AI, this bug appeared when a new feature was added to allow the user to input multiple inputs without confusing the AI.
A new warning dialog lets the user know if the application is being hosted when switching to public chat room chat mode.
Small adjustments have been made to the text displaying who is an AI bot in the public chat room, the colour of the text sometimes clashed with different interface themes, this has now been adjusted to fit the interface theme colours
Slight changes have been made to the public chat when viewing the other people chatting to the AI bot to make it clearer who is chatting to who.
A few things have been added and changed in the question detection part of the AI Core which will allow better detection of questions, it is recommend to restore the default AI Core in this version.
If you are a subscriber using a custom core, it is recommended that you back up your AI Core file before doing this, there are further plans to improve the question detection and self detection system in the future.
...Asking more questions about questions that maybe asked!...
Version Update 5.45
Core AI & brain-file update
Version: 5.45
Previous Versions:
5.44 - 5.43 - 5.42 - 5.41 - 5.40
5.37 - 5.38 - 5.39 - 5.36 - 5.35
5.34 - 5.33 - 5.32 - 5.31 - 5.30
5.29 - 5.28 - 5.27 - 5.26 - 5.25
...Click here to download the latest version of AIChat...
Version 5.45 Summery
Features added or changed:
- Improved the speed and accuracy of the "Check Brain" function.
- Improved the speed of updating save file format.
- Stray files and question folders have been removed from "View Mode".
- Added the themes "Midnight" and "Chocolate".
- The "Chatter" tab has been renamed to "Users".
- Changed the "Add Chatter" button "Add Bot".
- Slightly rearranged the control panel buttons.
- Bug Fixed: Topic change retention data updates quicker.
- Bug Fixed: Retention data now fully clears when using "Force Say".
Subscriber features added or changed:
- Improved "Load And Merge" speed and accuracy.
- Added the themes "Coffee" and "Coffee Cream".
- Bug Fixed: "Average Mode" for "Load And Merge" has been fixed.
- Bug Fixed: Clearing "Search" now does not search everything.
Version 5.45 Details
Stray files and question folders have been removed from "View Mode" as they begin to lag and freeze the window with larger brain files. The search pattern for stray files and question folders is an intensive process but it is not an unwanted feature, the ability to search these things will likely be reintroduced in a different way.
The "Check Brain" function found in the "Info" tab is used for fixing general brain file errors, it is automatically used in other aspects of the software, when an older save file is loaded this updates the save file format and is the major element in merging multiple brain files together.
Vastly improving "Check Brain" in speed will be noticed by anyone who has undergone a save file format update or who has merged two brain files into one. Improving these systems was in the area of a major bug, now fixed, "Average Mode" for "Load And Merge" is now working!
When merging two brain files "Adding Mode" simply added up the two matching file response counts together when merging the statistics, where "Average Mode" finds the middle between the two statistics.
...Updating the updater so it updates faster!...
Version Update 5.44
Core AI & brain-file update
Version: 5.44
Previous Versions:
5.43 - 5.42 - 5.41 - 5.40
5.37 - 5.38 - 5.39 - 5.36 - 5.35
5.34 - 5.33 - 5.32 - 5.31 - 5.30
5.29 - 5.28 - 5.27 - 5.26 - 5.25
5.24 - 5.23 - 5.22 - 5.21 - 5.20
...Click here to download the latest version of AIChat...
Version 5.44 Summery
Features added or changed:
- The exp needed for intelligence level in database details has been made easier
- Bug Fixed: Infinity loop has been fixed for multiple inputs
- Bug Fixed: The topic change option is working again
Version 5.44 Details
The new intelligence level system found in by pressing database detail button has been made easier as it get harder and harder to add new content over repeating old content.
A massive bug which slowed down chatting to the artificial intelligence on faster systems and crashed the artificial intelligence on slower system when multiple things were inputted has been fixed, this was made to preventing the artificial intelligence from getting confused between conversation data which was much further back, further improvement will be made to this feature in the future for better conversation retention.
A small bug which stopped the change of topic when no data was found multiple time in a row has been fixed.
...Not in a loop!...
Version Update 5.43
Core AI & brain-file update
Version: 5.43
Previous Versions:
5.42 - 5.41 - 5.40
5.37 - 5.38 - 5.39 - 5.36 - 5.35
5.34 - 5.33 - 5.32 - 5.31 - 5.30
5.29 - 5.28 - 5.27 - 5.26 - 5.25
5.24 - 5.23 - 5.22 - 5.21 - 5.20
...Click here to download the latest version of AIChat...
Version 5.43 Summery
Features added or changed:
- Better server admin checking procedures!
- Added an intelligence level to the database details button.
- Added a "Clear brain with biases" button to the "Bias" menu tab.
- Bug Fixed: The new folder "button" with the new statistics!
Subscriber features added or changed:
- Added a compress brain button to aid with game script efficiency.
Version 5.43 Details
The right click "add input folder" has been the quickest way to make a new folder in the brain tab, and has been up to date thus far, the add new folder "button" at the control panel at the top of many menus has not, and now been updated with the same functions! The main statistic missing from using this buttons was the creation date of new folders, files in the brain tab are the main output when processing such data, this should not effect overall performance, just data keeping for moderation and admin controls.
Total conversation from the old admin system to the new admin system is complete, with hopefully little interruptions from earlier version to this version, in the previous versions multiple admins could be in total charge of the server options, now that has been narrowed down to just the owner, just one BYOND username can now be admin.
Currently the word bias system blocks sentences containing banned words from being said by the AI, but the brain file still stores this data from the person who said it to it, you can now remove this data completely with a simple button very quickly.
A new ability to prune out nasty content from the brain based on banned words in the "Bias" tab has been added, this new button will scan the AI brain for banned words and then delete files and folders containing these words.
For subscribers that create, to help with the efficiency of the publicly available game script, a new button in the "Info" tab allows users to compress their created characters AI brain ready for any game or application, this will remove all the empty folders, ignored or a bias banned word and set every response remaining to 1 weight.
The new compress button provides a AI brain ready to be saved that is quicker to convert by the scripts first time converter, and also processes responses faster in any created game or application using the script.
After the brain has been converted, with "Learning Mode" off in the "Control" tab and "Use Question Detection" off in the "Options" tab, this can emulate what it would be like to talk to the AI using the script in a game or application.
The download now comes with a basic Chinese brain and Chinese AICore.
...More Power!...
Version Update 5.42
Core AI & brain-file update
Version: 5.42
Previous Versions:
5.41 - 5.40
5.37 - 5.38 - 5.39 - 5.36 - 5.35
5.34 - 5.33 - 5.32 - 5.31 - 5.30
5.29 - 5.28 - 5.27 - 5.26 - 5.25
5.24 - 5.23 - 5.22 - 5.21 - 5.20
...Click here to download the latest version of AIChat...
Version 5.42 Summery
Features added or changed:
- Further security improvements.
- New saving and loading for backing up AICore bias data.
- New load AICore button has been added.
- Changed a computer term to a psychological term, "Back-Tracing" is now "Retention".
- Added the display of what is consider old data to the option.
- Changes have been made to the default ignore old data option.
- Bug Fixed: Fixed a major word bias system bug.
- Bug Fixed: Loading user statistics.
- Bug Fixed: Deleting a file or folder error has been fixed.
Subscriber features added or changed:
- The AICore can be backed up and saved.
Version 5.42 Details
Back tracing has now been renamed to retention as it is designed to add a multiplier to parts of the conversation further back in the past, each file contains data which can store further back in the conversation and is used to link what was previously said to add enhanced retention in conversation, this allows the AI to be less random.
Added details in the options about when data is considered redundant and should be ignored as old data, the default setting for this was 5 years and has now the default is 10 years, this default maybe increased further in future versions, but can still be manually set by the users.
Since the security update of adding a new Admin tab to the menu, user statistics on how much an individual has chatted to the AI and how much the AI has learned from a user on each individual server has been restored.
A major bug has been found in the word bias system and has been fixed, this was noticed when a banned word was outputted by the AI, the change may effect small words being detected within bigger words but will allow for the bias system to be used by different languages in the future that do not use the space bar, further adjustments in the future updates will address this issue.
Further security improvements. In the new admin menu, mods can no longer add or remove mods, only admins can add and remove mods. Other improvements have been made to the new admin systems, implementing the old system with the new system, editing and deleting options for each level of user on a server has been fixed.
In the last version some moderators had some access to some admin only options, this has been corrected. The multiple admin system has now been fully phased out for the new server admin lock, locking one user as the server admin.
More stability and ease of access has been added with the option to backup AICore and AICore bias data with a new loading and saving function, this is locked to the server admins only and will remove the need to manually move files to the application folder.
The save AICore function is a subscriber feature as only subscribers have access to the ability to edit the artificial intelligence core, anyone can load these files.
Added an few example files which can be loaded and tested, including:
defaultaicore.aicore - alphachinesecore.aicore - exampleaicorebias.bias
Recent default brain file details:
Folders: 2567 - Files: 3763 - 6330 total files and folders
Responses: 8709 total file weight
Average Files Per Folder: 1.46591
Links: 3440 files linking to a folder.
Retention Links: 1880 in files retaining conversation.
Stray Links: 398 folders with no link to it.
Empty Folders: 821 awaiting a reply.
Total Characters: 260474.
...Something, something, universe something, Babel fish!...
Version Update 5.41
Core AI & brain-file update
Version: 5.41
Previous Versions:
5.40 - 5.37/5.38/5.39 - 5.36 - 5.35
5.34 - 5.33 - 5.32 - 5.31 - 5.30
5.29 - 5.28 - 5.27 - 5.26 - 5.25
5.24 - 5.23 - 5.22 - 5.21 - 5.20
...Click here to download the latest version of AIChat...
Version 5.41 Summery
Features added or changed:
- Changed the debugging text.
- Recycle bin saving has now been changed to be stored in the main brain file.
- A new "Admin" tab has been added to aid server hosts with moderation and banning.
- A whole new rewrite of the banning system.
- Removed right click mod controls in the "Chatters" menu, it has all been moved to "Admin".
- Server admin now locks to the first user and is saved as "/settings/serveradmin.sav".
- Added the ability to remove the server IP from being displayed in the "Info" tab.
- Changed the default theme to "Paper" the windows gray default.
- Bug Fixed: The cancel button for the themes button no longer sets the default theme.
- Bug Fixed: A mute person error has been removed.
- Bug Fixed: An AI chatting to an AI will no longer leak into other view modes.
Version 5.41 Details
When adding an AI chatter to make the AI to chat to itself using the loaded brain file, the AI will no longer leak into other chat room modes when changing the conversation topic.
There is an error that a person was muted when talking in public chat room mode and not chatting to the AI, this has been removed.
The debugging messages have been changed a little to keep bug reporting easier.
Recycle bin no long will be stored as its own files and will be stored in the main brain file, load and merge will not load the recycle bin of the second file.
New host controls have been added for people with the "Admin" access level, this will allow admins to ban people from the server, mute people in public chat mode, ban a user from teaching, ban users from editing and deleting file.
The changes in the banning system allow admins to ban users even if they are offline by entering their username in host menu and saves space when it is saving data, rather than saving in individual user save files, it saves in to its own unique file in the settings folder.
Added the ability to hide the servers IP address from general users, admins can always see their IP address, the default setting does not show the server IP to the public.
Normally the admin of a hosting server always had to log in, now the admin name gets saved for every other time, so the server host only needs to access the server once to lock it to them, and each server should now only have one admin.
This update generally improves server security, controls improve safety of all server hosts from destructive individuals.
...More control!...
Version Update 5.40
Core AI & brain-file update
Version: 5.40
Previous Versions:
5.37/5.38/5.39 - 5.36 - 5.35
5.34 - 5.33 - 5.32 - 5.31 - 5.30
5.29 - 5.28 - 5.27 - 5.26 - 5.25
5.24 - 5.23 - 5.22 - 5.21 - 5.20
...Click here to download the latest version of AIChat...
Version 5.40 Summery
Features added or changed:
- Added a view all mode for viewing others in AI chat mode, while in chatroom mode.
- Added a warning to the beta self recognition system option.
- Added an End User Licence Agreement button to the info tab
- Bug Fixed: Further fixed compatibility issues with the stolen alpha version.
- Bug Fixed: A muting error has been fixed for chat to AI mode.
Version 5.40 Details
This version further fixes the issues with freezing while loading alpha files along with other major loading bugs.
People can now see others chatting to the AI while in public chat mode, this will allow moderators and admins to better see who is inputting what to the AI.
A warning system has been added to the beta self recognition system as it is not complete and edits what is inputted in the brain file.
...Peek-a-boo!...
Version Update 5.37/5.38/5.39
Core AI & brain-file update
Version: 5.37/5.38/5.39
Previous Versions:
5.36 - 5.355.34 - 5.33 - 5.32 - 5.31 - 5.30
5.29 - 5.28 - 5.27 - 5.26 - 5.25
5.24 - 5.23 - 5.22 - 5.21 - 5.20
...Click here to download the latest version of AIChat...
Version 5.37/5.38/5.39 Summery
Features added or changed:
- Added options to hide the bias tab from users.
- A maximum character input was added.
- Extra menu hiding for admins to run faster server when hosting and using.
- Bug Fixed: Bias menu bug fixes.
- Bug Fixed: Fixed compatibility issues with the stolen alpha version.
Version 5.37/5.38/5.39 Details
This version a general update to fix back compatibility bugs and help speed up using and hosting AdamAIChat.
Further fixes between the modern version software and the stolen alpha version brain files have been applied to allow older users to bring their brain file up to date for free.
* Note that the older alpha brain files may freeze the application while they are being converted, this is because the file size is a large file, please be patient and leave it to run, it will unfreeze again. There is currently no loading screen for this process.
Options to hide the word biases have been applied, this will allow server hosts to hide the rude words they might not want users to see.
A maximum character input has been added to limit the amount of data in one sentence from clogging up the brain files, this can be set by the admin of the server in the options.
An option has been added to hide the brain and bias tab to allow for hosts to run a server at a faster speed, this is also found in the options.
...A blast from the past!...
Going International
正在走向国际，中文是世界上使用最广泛的语言！（英语是第二大语言！）
我刚刚发布了一个针对中文的实验性人工智能核心！
这不会改变按钮，但你仍然可以对它说话并教它！
请访问 AdamAIChat.com 的下载页面来帮助测试！
替换“settings”文件夹中的AICoreMain.AICore
尽管默认的大脑文件是英文的，但这只是为 AdamAIChat 制作中文大脑文件的第一步。
测试中国人工智能新核心正式服务器上线！
我已经注意到字体大小可能需要变大！
计划让用户能够在 AdamAIChat 的未来版本中调整字体大小。
由于英语和中文在语法上有很大不同，可能需要对代码进行一些更改才能完全实现该文件的所有功能。
你仍然可以在中文服务器上说英语，它有一个英文大脑文件，但它会对未知句子给出中文响应。
Official server online for testing the new Chinese artificial intelligence core!
I have already noticed the font size might need to be made bigger!
There are plans for the users to able to adjust font size in future version of AdamAIChat.
Some changes may have to be made to the code to fully implement all the features of this file, as English and Chinese are very different in grammar.
You will be able to still speak English on the Chinese server, it has an English brain file, but it will give Chinese responses for unknown sentences.
I have just released an experimental artificial intelligence core for Chinese!
This will not change the buttons yet, but you can still speak to it and teach it!
Visit the download page of AdamAIChat.com to help test it out!
Replace the AICoreMain.AICore in the settings folder.
Even though the default brain file is in English, this is one step towards making Chinese brain files for AdamAIChat.
Going international with the most spoken language in the world, Chinese! (English is the second most spoken language!)
在过去的 6 个月中，统计数据显示 55% 的网站访问者使用普通话，即简体中文。
45% 的网站访问者来自香港，10% 的网站访问者来自新加坡，而只有 25% 的人以英语为母语。大约 25% 的人使用世界各地的其他语言。
如此多的网站浏览量令人惊讶，因为 AdamAIChat 最初是为英语人士构建的，并向英语受众做广告。
原本的计划是通过切换一个简单的人工智能核心文件来创建一个灵活的通用语言系统，但没想到很早就这么灵活了。
据统计，来自香港的浏览量最多，AdamAIChat将在未来版本中调整针对中国人的功能！
In the last 6 months, statistics show that 55% of website visitors speak Mandarin, also known as Simplified Chinese.
45% of website visitors came from Hong Kong and 10% of wensite visitors came from Singapore, while only 25% were English speaking as their native language. Roughly 25% speak other languages around the world.
This amount of website views was surprising as AdamAIChat was originally constructed for English speaking and advertised to and English speaking audience.
The original plans was to create a universal language system that is flexible by switching a simple artificial intelligence core file, but I did not expect it to be this flexible early on.
With statistics showing the most view from Hong Kong, AdamAIChat will be now be adjusting features for the Chinese people in future version!
这不会改变按钮，但你仍然可以对它说话并教它！
请访问 AdamAIChat.com 的下载页面来帮助测试！
替换“settings”文件夹中的AICoreMain.AICore
尽管默认的大脑文件是英文的，但这只是为 AdamAIChat 制作中文大脑文件的第一步。
测试中国人工智能新核心正式服务器上线！
我已经注意到字体大小可能需要变大！
计划让用户能够在 AdamAIChat 的未来版本中调整字体大小。
由于英语和中文在语法上有很大不同，可能需要对代码进行一些更改才能完全实现该文件的所有功能。
你仍然可以在中文服务器上说英语，它有一个英文大脑文件，但它会对未知句子给出中文响应。
Official server online for testing the new Chinese artificial intelligence core!
I have already noticed the font size might need to be made bigger!
There are plans for the users to able to adjust font size in future version of AdamAIChat.
Some changes may have to be made to the code to fully implement all the features of this file, as English and Chinese are very different in grammar.
You will be able to still speak English on the Chinese server, it has an English brain file, but it will give Chinese responses for unknown sentences.
I have just released an experimental artificial intelligence core for Chinese!
This will not change the buttons yet, but you can still speak to it and teach it!
Visit the download page of AdamAIChat.com to help test it out!
Replace the AICoreMain.AICore in the settings folder.
Even though the default brain file is in English, this is one step towards making Chinese brain files for AdamAIChat.
注意：有关 AdamAIChat 的建议和错误修复请发送至 zelldot@gmail.com 以获取普通话/简体中文字体。BYOND 论坛与普通话/简体中文字体不兼容。
经过进一步测试，BYOND 仅使用拉丁字母中的字符，这会影响用户名、服务器状态和论坛帖子。
不需要 BYOND 用户名来运行 AdamAIChat，因为每个人都可以以访客身份登录，而无需注册 BYOND 服务。
Note: Suggestions and bug fixes for AdamAIChat should be sent to zelldot@gmail.com for Mandarin / Simplified Chinese font. The BYOND forum is incompatible with the Mandarin / Simplified Chinese font.
On further testing, BYOND uses only characters from the latin alphabet, this affects usernames, server status and forum posts.
A BYOND username is not needed to run AdamAIChat, as everyone can sign in as a guest without the need for signing up to the BYOND service.
Psychology And Neurology In AI
Click Here To Download AdamAIChat!
There is hardly any psychology of machines when it comes to neural networking architecture, neural networks are set nodes that are adjusted in value when trained, but new nodes are never created or destroyed unlike neurons in the brain, this makes neural networks less flexible.
Photo by Ilya Grigorik, Edited by Adam Watson
The way which data is structured in the current models of artificial intelligence is a static model unlike AdamAIChat, current models have a set grid of nodes which are adjusted numerically when trained to adjust the output bases on inputs.
In AdamAIChat the artificial intelligence element adds nodes creating a tree like grid with webs that connect higher points of the tree back on itself to lower points in the tree.
Certain features also allow for individual neuron to die, just like a human brain, the lack of stimulus causes a neuron to die, and with lack of a need for that neuron, older information can be lost.
In AdamAIChat the user has ability to set a redundant information time span on data which is not brought up over a long period of time, this feature allows for the death of older information which is not longer relevant or related to the current time.
As each individual node is created and olders nodes die off, this creates a more realistic structure of a brain than a static node model.
With each node being created as need, the neural network of AdamAIChat has much more flexibility when being trained, allowing for more user control over each individual node compared to the traditional idea of a neural network which is like a black box, where the underlying workings are hidden.
Due to the complex structure of the way data is constructed in AdamAIChat, other psychological elements have been added over the many years of AdamAIChat being made public.
First information bias is one the major features which is a very big human trait, the first thing that is said about a topic tends to a higher value over newer information.
With a mixture of redundant information, recent information biases and new information biases, the data which is AdamAIChat outputs if relevant and up to date.
Other psychological biases such as word based biases can also adjust the outputs of the artificial intelligence to create different personalities giving each individual copy of AdamAIChat, this gives the artificial intelligence likes and dislikes towards different types of conversations even with different brain files.
These word base bias adjustments can change the output of each conversation towards certain topics and move the conversation away from other topics.
The overall structure is a much larger enhancement to the original concept of neural networking with access to human like psychology compared to most of the original types of neural network design.
The End Of Nostalgia
With version 5 reaching its 4th year of its development, the nostalgia version of AdamAIChat will be decommissioned as further development cease to exist for version 1. Version 5 of AdamAIChat has all the original systems to emulate the nostalgia version but is much more efficient at compressing brain files from 24 gigabytes to 3 gigabytes.
 |
| Image by Ilya Grigorik |
Currently the game script run on the version 1 architect but in the next few month development of the version 5 architect of the game script will begin making it a lot faster when implemented into a game on BYOND.
The last versions of the game script and nostalgia reverts any version 5 script to version 1 and version 5 can convert any brain file to the latest version since version 1, with the upgrade to of the game script conversion time will greatly be decreased as some report that it can take up to an hour and a half to convert large brain files.
A nostalgia AICore file will be made available for that original feel of nostalgia without the original bugs and errors of nostalgia, with the latest user interface adding more security and control than ever before!
With many more features over the original version including subject retention, question detection system, psychological bias controls, word based bias controls including banning words, date based redundant information settings, easier to edit brain options, and easy hosting and moderator settings, the latest version 5 of AdamAIChat leaves version 1 nostalgia far behind.
Version 5 of AdamAIChat has all the original feature and functions of version 1 nostalgia and is built upon version 1, so no function is lost and every update since has added new features.
As more people tend to want the latest version, ignoring the old version, this is the best way forward!
Subscribe to:
Posts (Atom)