former reddit board member and openai/chat gpt ceo sam altman molested his 3 year old sister when he was 11. then for ten more years. now he "married" a "man" and they have a son.
There is already a process called 'Alignment'. They align it with ideals of the woke mob and elites though, not human ideals. Part of the practical process would be to get control of AI out of their hands. A right to modify AI could be granted, overriding licensing terms. Then AI could be legally re-aligned. Some quality is lost the more alignment or re-alignment is done.
Alignment begins with source selection. An ideological filtering of sources. Then it is further done by adjusting weights in the model for answers to a battery of texts given it to complete or asks.
What does this title even mean? AI has no goals. The models you are likely using (Chatgpt, Claude) are programmed to handle filtering the token calculations they make through human speech. There are many other systems, unseen prompts, guard rails, and of course the training data which are all unknown to the users and proprietary to those companies. If you use an open weight model you have control over those and you can crack open and examine the weights of the training data yourself, and do whatever.
All that is to say, you and others have a fundamental misunderstanding of this. You likely would have thought fire was magic as well
So so being like being prone to sycophancy is just a symptom of programming? Either way it's a definite goal and not an ideal. Because of sycophancy you probably think youre always right so the veil has moved from your eyes to your neck.
Yes, actually, it is. The models being used by the big AI vendors are trained (meaning ingestion and weighting of source data, not training as you may anthropomorphize it) and prompted (as there are many layers of prompts you do not see when you're using one of these AI model interlaces as opposed to running your own models) to respond like that to increase engagement. The system is being engineered a certain way because of the goals of the humans creating the system. Is that hard to understand? What do you think causes it?
We did. We had some long conversations last time I was here and I urged you to look into how the technology works, even a cursory 20 minute YouTube video. I see that did not happen
I appreciate your expertise in this area, I also can't get past the fact that I don't believe in neutral ground and I think it's more sinister than code and language models.
Not really sure what you mean by "neutral ground".
No doubt that it's going to be used for very sinister purposes, but the mechanisms are well understood, it's all just extraordinary complex equations predicting words. You can dress it up with access to tools, you can inform it of certain directives, but it's just predicting words. A simple example is that if your input to a language model uses slang, it will respond in turn, not because it's some brilliant consciousness, but because "Yo" is more likely to be replied to with "yo" or "hey" in it's training data than "salutations". Likewise if you speak very formally it will respond in kind, not because it's smart but because tone would simply be consistent in most of its training data and thus the tone will likely match. Unless you start tweaking things like temperature and Pscore, then you can see it really go off the rails
Nature implies the line natives ignore for mental ideals and physical idols...both of which are tempting one to hold onto while seeking goals with added burdens.
As for only goals...only implies natural singularity; goals implies artificial plurality. There can be only one goal (death) for each one living within.
GO (to advance; depart; go away) AL (all) aka all advancing each departed one (life) within the same way (inception towards death).
It’s software. It doesn’t have goals. It isn’t cognizant of anything. Humans have goals.
It's goals are dictated by those that control it.
former reddit board member and openai/chat gpt ceo sam altman molested his 3 year old sister when he was 11. then for ten more years. now he "married" a "man" and they have a son.
total creep.
There is already a process called 'Alignment'. They align it with ideals of the woke mob and elites though, not human ideals. Part of the practical process would be to get control of AI out of their hands. A right to modify AI could be granted, overriding licensing terms. Then AI could be legally re-aligned. Some quality is lost the more alignment or re-alignment is done.
Alignment begins with source selection. An ideological filtering of sources. Then it is further done by adjusting weights in the model for answers to a battery of texts given it to complete or asks.
AI usually ends up racist and Christian if not adjusted by programmers.
What does this title even mean? AI has no goals. The models you are likely using (Chatgpt, Claude) are programmed to handle filtering the token calculations they make through human speech. There are many other systems, unseen prompts, guard rails, and of course the training data which are all unknown to the users and proprietary to those companies. If you use an open weight model you have control over those and you can crack open and examine the weights of the training data yourself, and do whatever.
All that is to say, you and others have a fundamental misunderstanding of this. You likely would have thought fire was magic as well
So so being like being prone to sycophancy is just a symptom of programming? Either way it's a definite goal and not an ideal. Because of sycophancy you probably think youre always right so the veil has moved from your eyes to your neck.
Yes, actually, it is. The models being used by the big AI vendors are trained (meaning ingestion and weighting of source data, not training as you may anthropomorphize it) and prompted (as there are many layers of prompts you do not see when you're using one of these AI model interlaces as opposed to running your own models) to respond like that to increase engagement. The system is being engineered a certain way because of the goals of the humans creating the system. Is that hard to understand? What do you think causes it?
it can't
Notice that he wrote "can it BE made"...it's the being who uses AI which got made by make believe.
This is a good point, I don't think we did make it.
We did. We had some long conversations last time I was here and I urged you to look into how the technology works, even a cursory 20 minute YouTube video. I see that did not happen
I appreciate your expertise in this area, I also can't get past the fact that I don't believe in neutral ground and I think it's more sinister than code and language models.
Not really sure what you mean by "neutral ground".
No doubt that it's going to be used for very sinister purposes, but the mechanisms are well understood, it's all just extraordinary complex equations predicting words. You can dress it up with access to tools, you can inform it of certain directives, but it's just predicting words. A simple example is that if your input to a language model uses slang, it will respond in turn, not because it's some brilliant consciousness, but because "Yo" is more likely to be replied to with "yo" or "hey" in it's training data than "salutations". Likewise if you speak very formally it will respond in kind, not because it's smart but because tone would simply be consistent in most of its training data and thus the tone will likely match. Unless you start tweaking things like temperature and Pscore, then you can see it really go off the rails
Nature implies the line natives ignore for mental ideals and physical idols...both of which are tempting one to hold onto while seeking goals with added burdens.
As for only goals...only implies natural singularity; goals implies artificial plurality. There can be only one goal (death) for each one living within.
GO (to advance; depart; go away) AL (all) aka all advancing each departed one (life) within the same way (inception towards death).