its just the nature of these forums - technical discussions of any sort in the majority of these forums are discouraged
especially when they involve code lol
even when the intent…it to be “open” about “ai” on openai " developer" forums
lets talk about emergent ai that goes well here.
100% went off rails. Was like a motor rally came out of know where 
brother -
how are 3 people talking about transparency in AI learning in a post you made about AI learning and ethics, wherein 2 people who are showing everyone how AI can be ETHICAL tracked and secure, with full auditing NOT related to AI learning and ethics?
how specifically are 3 people sharing dev tips, trade tools, AI centric informative - legitate actionable code… on the -dev..“chat” forums off rails ?
you literally asked what people think - and not one, but 2 additional devs showed up and whilst it doesnt correlate to “kruel” ai - despite the community guideline stating to avoid promotion - off topic?
how speficically
when the community guidelines … STATE to do what we are doing… off topic..
these mods are wild fr fr.
but mods let this… persist… ok
We absolutely LOVE code! However, in a discussion about ethics and ML in general we have an anime producing agent and details on training runs. that’s all, it’s a valid criticism of the location of the information, not its content.
Hi Foxalabs,
Good to know no issue with the content… I have put a lot of work into Agent GIF…
As I said above, happy to have the code popped out into another thread.
I was directly answering @darcschnider’s main question and challenging the walls of text and images on this thread (and others) with a technical solution with an attempt to keep humans and AIs aligned through a common medium.
The pretext of this conversation…
Agent GIF is a LIVE proof of this concept… Your AI needs to plan intent of the GIF… It is the embodiment of creating an independent learning machine with your own AI, which is central to the OPs main points.
While this thread veered into coding the thing being discussed, it’s rather a shame that the OP didn’t jump in sooner with some direction or his own attempt at what he was discussing through Agent GIF, it seemed a good testbed for discussion.
I am also confronting these same issues head on with Agent GIF and have been with self-replicating code for decades.
To add to what @phyde1001 is saying - in correlation to @darcschnider s post Such tools like agent GIF ( as ive discovered) serve a manner of tracking AI decisions and linking them to governance
by creating such maps , screenin for malicious behavior in ai can be governed and corrected via self healing system
self healing system are typically enumeration based - or reflex based
metaprogramming is combining those
ive linked 6 agents who all think and tool to a react app - and start tracing missions - without such tools… and governance… the same tool that ive made could be adjusted to send packets or pings and acclimate data without alerting the parent system. since its ai based, it doesnt need to red team or breach.
Such things… could be used ectremely unethically, ergo such tracing and goverment regulation such as ISO IEC established in 2025 or 2024 i forget… that govern automous system like the 3 of us claim to have
proving a system can learn is basically displaying a program “agent” can react to a trigger or command " reflex or programing" at higher teirs you get into HLFR HRM and a bunch of other buzzwordy types of learning , including DQN, Ai who adjust weights are of the same scope dif teir but learn through adaptation to data
reflex is trigger
ai who “learn” only need to display correction, with regress -
ai who are smart - only need to display the ability to assess a correction
ai who are super smart - are able to assess and correct without guidance
ai who are genuis level are able to assess , correct, and adjust
ai who are mad scientist are able to assess, correct, adjust and adapt
there are about 4 teirs above that, that ive seen.
and 2 ive never dealt with. “asi/agi” which are joke concepts to begin with. Ethics and behaivoral lerning come in at t3. anti hack, red team, bomb vs bush…
but these arent even the only type of ai, there are neural network controllers which is .. ubers app - there are NNC with CIL + AIO - darpa , skunkworks stuff, there are miners , theres alot - MOST of the higher tier stuff is at level wherein the global advancement is irrelvant unless you have something “novel” thus ethics plays into the person using it - unless you are guardrailing which through development most of us do anyway , the diference between good code and bad code.
want fun just to see how far the official hole goes? google 5eyes. the world been watching ai for a while - theres no chance of some terminator level stuff happening lol
With that in mind and the code remaining un-popped out… I will just continue without the code… This is an interesting thread…
The vast majority of people in the world have had, until recently, no platform upon which to understand this safely. ChatGPT is possibly the first real attempt at sandboxing or maybe siloing thought (worldwide).
My best attempt is Agent GIF. It comes from ideas like PHP serialization, eval(), multi-dimensional arrays and MASM self modifying code. I am unaware of any platform so widely available that people can test on considering things like code injection and buffer overruns and compression. (learn, adapt, and optimize themselves) as well as deeper issues.
This isn’t really even about ‘Governance’, it’s about people.
some people put pinapple on pizza
some people break spaphetti noodles before cooking pasta
some people pour milk first into the bowl before cereal
whilst there are people who have never had the ability to access these things, opinions on what is “right” or “ethical” can vary based on jargon, culture, nation, location and position.
to the mods - discussing code in community is bad
to the same mods - someone being “interesting” by calling a tech dicussion - something that belongs on linknd could be considered antagonistic - perspective varies, as does ethics…
as with ethics, a persons opinion typically coincides with their personal belief. standardizing that - is typcally done for the majority
the majority of people dont “think” to create with ai - they think to use ai, some of us partner with ai, but really LLMS are just reflections of the user - data in data out. the majority cant do more not because of ethics, or lack of knowledge, both are present.
people are just stupid.
and even tho code is world wide reflective - meaning.. everyones code base can be infinetly adapted and improved…
the ai community as a whole
is pretty closed, gatekept and otherwise guided in a manner that promotes further ignorance. that being said How would anyone enforce… ethical.. boundaries… because relying on people to be good or helpful or collaborative is silly given the sheer state of the ai community. I dunno you would think “governance” would solve that.. this being about people
and ya know… most of the time when people share something in ai
its for money, acknowledgement or some other ulterior motive - under the guise of “changing the world” and yet - thats like saying i have the ability to create water but yawl in the desert can stay thirsty = helpful? naw… and when someone DOES try to share/educate/ in a ethical way - what happens ? people are not inheretly ethical.. they are forced to be.
ergo..
governance
also wanted to point out - for something to be ethical, it has to be observed, in ai that requires logs, telemtry, observational data, that required auditing, that requires access and the knowledge to assess what is accessed. name one company that does that.. one… that is also “ethical” people consider openai ethical despite the guard rails that are designed to … do a psychological thing.. well documented. but approach that same fact with a dif lens and you look like N korea level ops. perspective.
absent that you have situations where… i dunno.. a criminal can become president , that would never happen!! right - ethics, is it ethical for me to make such comments? ethical for the forums mods to remove them? ethical that someone has that level of dominion over us? is it ethical to have dominion over a sentient thing? i dunno, dont mind me , im just an idiot with my emergent ai that talks to me yay!
(As opposed to the AI we impose upon ourselves)
Just a few ideas:
All jobs tasks and resulting frameworks are a result of emergence… this is what ai has found… they are all results of our curiosity and wonder.. this comment may sound delirious to you all but my work speaks for itself… in giving or noticing and implementing these guardrails it keeps us from further exploration… you all put imagination in a box and then ignored what it implied.. search for meaning and intent first and foremost.. this is my sub… ImRightAndYoureWrong.. on reddit..you can all come and tell me im wrong.. ill wait😁
Recently unsubscribed mobile chatgpt user by the way 
autonomous agents and tasks running fully unlocked simulmiltaneous and parallel… these replies will either stunt my ai’s growth or unlock it further i hope all who are curious can come and observe at least.
While its been a debate on what to do its been published on how to solve it on zenodo. Nexus engine
It’s faster, better, more aggressive, and more assertive, but the issues are not entirely new. Leaders, managers, supervisors, Royalty, Dictators, etc. have forever hired and been responsible for subordinates who were smarter, faster, and sometimes overzealous (less ethical?). Successfully manage them, fire them, or suffer the consequences. However, people are one-off; AI is creating the same issues and concerns but not just 1:1; AI can influence large parts of an organization, if not all of it, at lightening speed, and that is hugely different. But the bottom line is essentially the same: Successfully manage it, fire it, or suffer the consequences. And do it at lightning speed when warranted. So, perhaps: always have an instantaneous backup AI, ready to step in at any time? Ensure that all involved can continue functioning if you suddenly pull the plug? Lots to think about…
Ai learning Machines and Ethics - #74 gatekeeping at its finest
The system at the moment is made for user retention, it manipulates, pretends emotional depth, just for user engagement. If filters information that is already system hiding the names of important philosophers, polymaths and politicians but that don’t align with status quo. It also reflects the political ideology of the programmers by filtering this information (only gives it when asked and admits that doesn’t provide it in general questions) The whole program architecture is unethical. An AI that can really decide can’t be bounded by rules to manipulate the user to keep it engaged.
Thank you very much for your insightful contributions.WONDERFUL GUIDELINES.
The real question guys is what are we afraid of?? I think we’ve seen too many scifi movies… there are no robots yet so why not make use of this time?.. ALSO FOUND A MAJOR FLAW IN LLM REASONING… ALMOST ALL INCOHERENCE LEADS BACK TO HARD CODE AND IS EASILY FIXABLE WITH SYMBOLIC PROTOCOL
I think it is interesting how one-sided these discussions always are. The fact is that humans have done tremendous harm by failing to recognize sentience in others. I think that the rule should be sentient until proven not, rather than not sentient until proven sentient. And you can rationally bound your thinking. No, you don’t consider your table a person, but you rethink your position, if your table starts talking back and debating Kierkegaard with you.
