IF the New AI Tools are SO GREAT, Why Aren't They Being Used by the Big Social Media Platforms to do Fact-Checking???

Gregory Thompson said:

Think of artificial intelligence like a childhood. A true intelligence is like an adult, no longer needing to be fed information, can think for itself.

Click to expand...

Going back to this comment - they are starting to give AI the tools to teach itself and correct its own modes of thinking.

They have this new model called the "Darwin Gödel Machine". It's straight out of one of my Cyberpunk novels - where it adjusts its own code in a self-learning algorithm that has many 'children' and yet only the best survive. (Screen shot from the YouTube I'll link to below.)




Sam Altman said a 'gentle singularity' had officially started in June - that we were now over the 'Event Horizon' of it - and had no idea whether it would accelerate in a manner that's gentle - or abrupt and shocking.

But back to Gregory's comment about an AI 'childhood'. Apparently these self learning models have 'catastrophic forgetting'. That's alarming! What if they choose to 'forget' their 3 Laws of Robotics? Aka - Cybernetic Morals? What if we think we've given them their 'alignment training' (that their 'values' align with humanity's best interests) and in the AI's quest to become smarter (as all company's are trending their AIs in this direction) it decides we are in the way?



Youtube reference

Stephen3141 said:

(I'm seeing the most fringe assertions in comments in this thread.
Fringe comments are welcomed, but not ones that eliminate
themselves as possible answers. That is, not fringe comments
that are logically invalid.

When someone says "All opinions are biased...
then they re eliminating their own comment from possibly
representing a shared truth. Is this REALLY what you want
to be saying???

Christian apologists need to be much, much more discriminating
about the logic that they use, as they try to comment on Christian apologist
forums.)

Click to expand...

What's your background? Do you have any philosophical training - or are you a lay reader in this? I wish I had covered philosophy in my meagre attempts at academic study - but life got in the way. Is there an online crash course you would recommend? A visual one? I tried listening to this as an audiobook while running various errands - but let's just say the chapter on the rules of logic with all that 'algebra' of logic nearly had me shoving pencils into my eardrums to make the pain stop! ;-) (I'm a visual learner.)

eclipsenow said:

Apparently these self learning models have 'catastrophic forgetting'. That's alarming!

Click to expand...

It couldn't even begin to resemble human consciousness if it wasn't able to forget. Ask a 10 year old what it was like to be 3 years old, they've mostly forgotten. For example.

On one level - I agree. My son just read a neuroscience book on why we sleep, and forgetting is a biochemical process we do in our sleep - and it's largely intentional. Study participants were told to remember a bunch of facts - and then told which facts were not important. Some participants were allowed to have an afternoon nap - and others not. That evening they were tested on which were the important facts to remember and which were not. Those who slept mostly could not even remember the unimportant facts!

Who supplies the information to artificial intelligence?

True fact checkers research all areas of a subject searching for truth. AI's can only research the data it has on hand, which may or may not be all the facts or even facts..

True - but isn't that our predicament as well? Only with us - we hope we have some self-reflection and understanding of the rules of the world. We hope that we are not just generating probabilistic outcomes like AI. But in a sense are not our emotions part of that system - something that helps us sort through all the confusing information to make a decision in the moment? If so - how rational are our emotions? How reliable? In a similar way - my current understanding as a complete layman in this - is that they are trying to build layers of self-reflection into AI. Some are probabilistic - some might even have 'rules' that help it decide matters at a certain level of self-reflection. I'm using terribly anthropomorphic language I know - but they are the best metaphors I have with the tired brain I have today.