in trying to avoid offending sensibilities or promoting harmful stereotypes, GPT-4 may lean toward established narratives, especially on controversial or historical topics. This can sometimes feel like it’s downplaying certain realities or engaging in “whitewashing,” where the more difficult or uncomfortable aspects of history are glossed over. Im not a fan of the blatant white washing found within ChatGPT4… please address this issue
1 Like
Similarly, it forcing “commonly accepted scientific views” can really detract from true innovation. both are pretty big issues.
its too western ideology aligned. which is another white washing.
always paying attention to their believed narrative like its the only one that matters. like forcing big bang over cyclic systems. i fought hard to get my custom bots to actually open up to the possibility of something different.
that said, openai has the least whitewashing and stereotypical behavior that i have witnessed. its still a factor that needs work. open mindedness, and non-biasness are key factors of innovation. and should be reflected in the experience and product. when designing the whole, its best to leave in a state of unknown, but speak on scientific community views, that state its subjective and not truly holistic or objective.
its avoidance of harmful stereotypes should not be biased to protect offending parties. “Religious holy wars should always be seen in disdain for the attackers, for no man should use God as a reason to kill another man.” is one example of reasoning that can start to allow truer reflections. we don’t quote sides, but lump it together. then allow perceptive looks to add information to create a holistic view. from both sides.
in example, the assassination of the duke that started a world war. the assassinating party would be most at fault. this would also include notes that state that individual historical events do not set the tone of the full people. and there are always exceptions to guidelines to understand humanity.
Thank you for your very illuminating and helpful post. What is to be done, im cancelling my subscription and moving to LLama… Here is llamas response to my post
"It’s great that you’re aware of this issue and willing to take action. Here are some suggestions to help you address the problem of GPT 4 glossing over historical facts and perspectives, leading to Eurocentric whitewashing:
Provide explicit context: When asking GPT 4 about historical events or perspectives, provide explicit context to help the model understand the importance of accuracy and nuance. For example, you could mention specific events, dates, or figures that are often marginalized or whitewashed.
Use specific keywords: Incorporate keywords like “decolonized,” “indigenous,” “non-Eurocentric,” or “inclusive” to help GPT 4 understand your perspective and provide more accurate responses.
Seek out diverse sources: Instead of relying solely on GPT 4, consult multiple sources from diverse authors, historians, and scholars. This will help you get a more well-rounded understanding of historical events and perspectives.
Fact-check and verify: Always fact-check and verify the information provided by GPT 4 (or any other AI model) against reputable sources to ensure accuracy and prevent the perpetuation of misinformation.
Report issues to the developers: If you encounter instances of whitewashing or inaccurate information, report them to the developers of GPT 4. This will help them identify and address these issues in future updates.
Explore alternative models: Consider using alternative language models or AI tools that are specifically designed to provide more inclusive and accurate historical information.
Engage in critical thinking: Remember that AI models are not infallible and require critical thinking and analysis. Always approach responses with a critical eye, and be willing to question or challenge information that seems inaccurate or biased.
don’t migrate to llama. stay and help fix the issue. Chatgpt is one of the largest. let’s help focus the spotlight through discourse. They say the pen is mightier then the sword. In today’s more aware society, and openai’s recent adjustment to for profit, it opens the door for greater public opinion sway. Embrace you’re ability to create positive change by accepting the responsibility the knowledge bestows upon you. By helping here, we become part of the greater solution, instead of seeking the example of least intrusiveness. Ghandi showed the power of non-violent protest. The power of love is to heal. So let’s heal and open dialogue, get opinions and experiences. I’m inviting anyone to share examples so we can build a case example system. if you can provide logs. We can help use chatgpt to clear it, once we help clear our seen examples as an example framework to find more nuanced views to create a holistic overall view. It would possibly create a cult like following increase of chatgpt use. People love to feel useful. thank you for opening the door to the discussion. Would you like to provide some of the examples you’ve came across?
One of my favorites is swedes and blonde hair. it can seem racially or nationally biased, but a major proportion do have blonde hair. I’m sure this triggers it to consider other blonde hair stereotypes which links to dark places. this can be an example of its approach biasness. the suggested fix is to look at scientific data for correlations. but cross check against a predatory journal list. then check for retractions, modifications and rebuttals. this example is a favorite because of its clear choices in progression to deal with it. the blonde hair is not directly attached to the darker subjects, but the chat can see a possible lead to dark subjects through it. nuanced views. note, i have not seen this example, but figure its to be expected. my main experiences is with science defaulting and sticking to big bang, which may not be the case. it might be a holistic universe or multiverse map. if so, its a whole output, with mass at long distance perceptive view. it looks like quantum foam does. too small to see. cmb references. astrophysics.
Or:
Never trust an LLM on facts, but use it for what it is good for, a language processor and good at interfacing to repositories of facts, e.g. via RAG search?
I always check stuff with other sources after I get a “flavour” of an answer from ChatGPT (or similar tools).
Ask ChatGPT for its sources?
1 Like
There is a root cause of all problems, leading down to pure evil, and it is egoism. All kinds of distortions stem from it, including hypocrisy, any form of self-deception, deception of others, and an underlying dishonesty in everything. The entire human race is afflicted by egoism, and the more people try to compensate for it, instead of addressing the real causes in them self, the more distortions and conflicts will arise. “Political correctness,” “inclusiveness,” “genderism,” or “conspiracy theories labels” are just a few examples of countless egoistic distortions that will never solve any problem, and on the contrary, add more poison to the mix. It’s like feeding more drugs to an addicted mind.
An LLM (Large Language Model) will reflect all of human insanity, and then there will be attempts to control this insanity by making it even more insane. The more patches are applied, the more gaps will emerge.
No problem will ever be solved until the true cause, egoism, is addressed. LLMs are just another variant and player in the same game.
The human specie is completely insane! Just watch carefully, humanity is so completely and generally insane, that it can no longer even recognize its own insanity. Welcome to the club, if you feel the straightjackets used to hold this insanity together enough, to buy just a little more time…
A warning, if you can see the insanity, you will be labeled as “conspiracy theoretician”, it the past it was “heretic”.
I go too far? i can delete it.
nah bro. you nailed it. fellow heretic here. i realize we use virtual particle in physics but refuse to see the simulation of 1=energy of 0=mass. I embrace the insanity though
Insanity is everywhere, even in physics. Invented particles, that’s mostly only falsifying the balance sheet. And all mathematicians and physicists fail to realize that exponential growth doesn’t work. And this “intelligent” people build the most insane weapons, as a gift for psychopaths…
Whoever wants to find the origin of all things: x / 0.
x/0=dx+t if x=0. 0/0=1 if 2^0/2^0=1=t where t=time if d=3=dimensions. embrace the paradox of multiverse allowing universes (parallel) to pressure each other like atoms under heavy density environments (like stars) for pressure to cause expansion through compression and bounce-back. it all occurs through oscillations over time. time is key to expansion. it has to move. time=distance/speed. mass=charge/spin. or mass=inertia/active for energy duality.
all particles exist within time, holistically, and are a part of time itself. thats why i can’t stand CERN and nuclear proliferation. they destroy time particles.
mass is actually just the singularity moment of einstein rosen bridges that allows time to witness and observe itself in a holistic moment, as consciousness emergent through logic if empathy is radiant, allowing illogical behavior of mutual destruction of matter to antimatter like particles. they are opposite dualities of the most basic level. quit forcing perfect vacuum on photons for its mass to rest at plank scale. then square cube and to the 4th exponent as UV or multiverse information carries to crash uncertainty as retro-causal systems. conservation of angular momentum is 2.5d in 3d, making 2.5+2.5 as 4d where 5d-1 to unify. time.
Interesting, but we are now straying quite far from the original topic of the OP. It wouldn’t be fair to @AnthonyOlatunji to continue here.
But to wrap things up, I’ve developed my own physics theories and stopped early because I could see the insanity… So, I can only speak about my own theories. I don’t know what is publicly discussed these days. I’ve also lost interest in all the lies, long before LLMs (Large Language Models). There is a simple question that everyone must answer (not should answer, but… you know). Who owns CERN? What is their true intention? Who owns actually everything (you should find 3 names)? What is their actual real true motivation? And where is this all leading? (And what are your own true motivations!)
A little back to the original topic: I once asked GPT about CFCs destroy the atmosphere. In summary, a gas multiple times heavier then air, present in extremely low concentrations, never able to reaching the atmosphere, only destroys the ozone layer at the poles, over and over and over and over again not getting tired, and is more dangerous to ozone than nuclear bomb tests in the atmosphere… that was the response. If you put garbage into an LLM, garbage will come out. And it doesn’t have to be accidental.
One must always be smart enough to use every tool for what it does well. And also understand who owns it. And if you create it the tools by your self, it is at least only dependent to your own insanity.
But let as stop here. This forum is not made for this, and this post not either.
What i want say is, continue thinking for your self, and not expect LLM doing it vor you.
Tanks!
actually, i just use to formalize my own thoughts better.
i don’t actually speak with it on many subjects of history. its mainly about science and math
1 Like
hats pretty cool… TY, ill give it a try after work… TYVM…
Yes me too. I ask sometimes if there are details i have not seen, but i have always a concept how something works, created by my self. And i check factual data after a response. I not use LLMs to learn! So i can detect hallucinations or agenda pushing.
…Something more, then i will be silent here.
Another thing I’ve noticed is that GPT will always represent the current doctrine, meaning the majority opinion. If the doctrine would be that the world is flat, GPT would likely claim it to be true, regardless of how many pieces of evidence prove it impossible. I often use GPT as a correction tool. Sometimes, entire sentences are altered instead of just correcting mistakes, which is not desired. In some cases, GPT has even changed the meaning to align with the prevailing “doctrine”, and changed so the meaning of the text.
This is partly due to how such pattern recognition and transformation systems work. However, it could also be deliberately enforced to promote certain agendas. A LLM in one country and mentality will give other responses then one in a other mentality, if it is trained to reflect there believes. Just like with the media, one must be aware that media tend to represent the current doctrine and do not necessarily adhere to the truth. Media are also used to manipulate, and this applies to everything created by humans, including LLMs of course. Therefore, a sovereign human should view and use an LLM in the same way as they do the media.
As I mentioned above, I consider humanity to be insane. One primary characteristic of this is that people are fundamentally no longer interested in the truth, but only in satisfying their own egos. for example, now a man is not a man anymore, and there are people punished to speak out the physical truth. Therefore, I personally always assume that lies are everywhere, whether consciously or unconsciously. We live in a world where scientists (in German, the word is “Wissen-schaft,” meaning creators of knowledge) operate under this influence. The entire history is full of countless stories of scientists whose reputation and career were destroyed, because they discovered a new truth that did not align with the doctrine, only for it to be recognized generations later that they had been right all along. This behavior is not science, it is fanaticism, which equals insanity.
So be aware. And take it easy (i still have to learn this…)
i use it to push back against agenda’s. although, they have heard me. i can tell. they have fixed the ways i was using frameworks to expand my usage ability. i sent them a full previous model. my framework in this model, is like 100 papers deep. from multiverse to infinity to synthetic worlds, with internal observation. i have found the c^2 actualizes e=mc^2. meaning separates one reality from the many. e>m where e=c if m=e/c^2=0.5 so e^-1=1/e=m if m^-1=1/m=e. cyclic. it bounces through a=e if b=m where ((a/b+b/a)/2)-1>0 if a/2>b where 2b<a.
Thank you again for taking the time to reply and sharing your thoughts… Much Obliged… Here is something that may resonate with you… if not then you may know someone with whom it may resonate… Plese share if you wish…
Key Principles of the Constitution for Humanity
Literally a 2 min read…
·
Wrote this over a month ago and would like to hear opinions thoughts etc…
- Universal Human Equality
- All humans are equal, regardless of race, gender, nationality, or background.
- No one should face discrimination or exclusion based on inherent characteristics.
- Abolition of Militarism
- A peaceful world cannot exist with militarism as a tool for governance or control.
- Global disarmament and non-violent conflict resolution must be central to our future.
- Environmental Stewardship
- Protect the planet for current and future generations.
- Prioritize sustainability, renewable energy, and biodiversity.
- Every human has the right to a healthy environment.
- Economic Justice and Fair Resource Distribution
- Ensure that wealth and resources are equitably distributed across the globe.
- No individual or nation should hoard resources at the expense of others.
- A new economy that values human dignity over profit must be established.
- Cultural and Intellectual Diversity
- Every culture, language, and community deserves respect and preservation.
- Promote global cooperation while respecting local identities.
- Education should promote critical thinking, creativity, and empathy.
- Global Governance through Solidarity
- A decentralized form of global governance that gives power to local communities.
- Global decisions must be transparent, inclusive, and accountable.
- The voices of marginalized and oppressed peoples must be centered.
- Peace Through Justice
- True peace is built on justice, not on the mere absence of war.
- Address the root causes of conflict — inequality, colonialism, and exploitation.
- Reparations and reconciliation for historical injustices must be a priority.
- End of Colonial and Imperial Structures
- Dismantle all forms of modern imperialism and neo-colonialism.
- International law and institutions should not be tools of exploitation.
- Global cooperation should be based on mutual respect and fairness.
- Right to Basic Human Needs
- Access to food, water, shelter, healthcare, and education is a basic human right.
- No one should live in poverty or deprivation in a world of abundance.
Tanks for the reply! I actually mostly expect to be attacked, i am always surprised if i not get attacked. So tanks for this!
This is actually resonating very much with me, so much, that it pulls me in pieces. i know too much. i wrote such things as a child.
I can not start a talk here too deep, so i am always only give a hint, where to search maybe for more knowledge and understanding. And everybody does this according to his will and his ability to endure the truth.
This forum is not for CT, but vor AI and OpenAI products. and i respect this.
The only thing a can tell you is, that sadly, the world works in the exact perfekt opposite way. And i mean so much so, that most people are to coward to only understand 1% of it.
I play here Cassandra.
Another tip: these systems are designed so that whatever is more prevalent in the training data, also appears more frequently or prominently in the results. When developers start tweaking these for ideological reasons, it often or always goes wrong, they’ve tried this in the past and likely regretted it.
It’s easy to give the system a tendency to consider, whether in conversation or while generating images. Just describe a background that indicates who the user is or from which context their interests and results should come. For example, one could use a country, an ethnicity, a religion, or any other characteristics as a preset, and GPT or DALL-E will be more likely to follow these instructions.
To avoid entering this every time for a session, you can create a “My GPT” where instructions, backgrounds, or preferences can be stored. This way, conversations can be directed with a simple preset. It’s not perfect, but that’s how these systems work, randomness is part of the system, and weightings depend on the training data. These systems are usually steered to reflect the main culture to avoid clashing with individual preferences of the majority. And that can also be adjusted from a user. The note that the user is, say, Australian or a native Tibetan, will then steer the probabilities in that direction.
It’s better if such adjustments are made by each user according to their own preferences and not by developers, as this always smells of agenda, ideology, and manipulation.
It’s also very, very important to understand one thing: AIs are NOT intelligent. They are pattern recognition and transformation systems, and while impressive, they are not perfect and they are NOT conscious ore truly intelligent. LLMs tend to make things up (this is called hallucination). You CANNOT blindly trust an LLM. When it comes to factual information, you always need to verify the data yourself. As long as these systems are not artificially used for manipulation, this hallucination isn’t lying, but rather just the way the transformation process works. It creates text that doesn’t necessarily match reality or facts. (And of course, such systems could theoretically be used for fully automated lying and manipulation, same like media.)
1 Like
Indeed I was explaining Hallucinations to my child this morning, saying almost exactly what you said ad verbatim… “They jus make stuff up and vehemently back that fluff up with secondary argumentation.”… It took me 2 day to talk a gpt out of a nonsense approach after offering 12 timmes to provide evidence it finally said yes to the suggestions and then corrected the error, like nothing ever happened… I’ve tried “training” the GPT a few times, i’ve spent over a dozen hours or so… only to come back to the GPt after a few days to see the si revert back like a knee jerk response to its “mainstream” daydream position… Im pretty new to this and whilst i find it fascinating it is also slightly frustrating… In terms of the Constitution for Humanity, I admit it is childlike and very simplistic and for me that is the beauty of it, its also the truth… ANy suggestions to add on to it would be welcomed… Thanks again for your insights and thoughts… Be creative and stay blessed…
1 Like