I have resisted using AI for a very long time. I have had a few plays today. The most obvious thing in the co-pilot back page summary of my dreams is that it mentions AI even though there is little occurrence in the fed data set.
AI has confirmation bias. So it is getting close to being human 😉
It is biased against quantum. {for this data set}
They don’t seem to like each other.
It seems according to the Mountain View Oracle there are already a lot of associations with the word dream and AI. Some of the more graphic as in “artistic” uses, allow you to generate some images, a “dream”. They are a bit cheesy and biased toward the younger human end of the spectrum with a bit of a thing for sci-fi androids and mobile ‘phone image filters. No sex pistols, anarchy or miners’ strike.
You can get AI dream interpretations on line. I wonder what that would make of Pharaoh’s dreams?
Would they all have gone hungry or called Just Eat?
I know bugger all about AI. This is my self-assessment. But the idea is that what you train it with predicates the outcome. The more average the input the more average the output. It has to, on mass feed, tend towards some kind of norm. {boring?}
Could there be an elite boffinacious AI? Maybe…
In principle it might be possible to train AI on dreams. First pass on the Mountain View Oracle turns up a number of these attempts.
In the limit true dreaming can be Soular intuition. Which has little to do with reason and data. Training AI to be genuinely as opposed to statistically intuitive might be tricky. With enough data some hunches can be inbuilt.
Could you train AI to dream like me?
In order to dream as opposed to think there has to be a different functionality, built in. A new layer, a new manifestation of AI. Not rational or comparative.
Is there such a thing as DI, dreaming intelligence?
Could you train AI to think like an inventor or entrepreneur?
If you could, there would be a lot of dosh in it…
If you think “in the box” you cannot train AI to think “out of the box” because you do not know how…
I had a little joke to myself. I was imagining myself trying to pitch a dreaming-AI concept to some technology VC. They asked for my mobile ‘phone number. I said that I don’t use ‘phones and that I am barely able to answer a call on one, on the rare occasions it rings. I really am unfamiliar with their use. I am cack handed with a mobile ‘phone. On his face I could see a huge WTF non sequitur.
“This dude cannot even use a cell ‘phone!! See ya…Bye…”
I am aware that my dream content has famous geezers like Penrose in, well respected academic institutions and things like a Perimeter Institute. Way back in my prior incarnation at the turn of this century I was aware of how important these things seem, in that context. It is an alternate reality to the one in which I live. It is/was a kind of vague dream-world of decades ago, an insubstantial echo in space-time.
I am half-tempted to prepare a huge document with all my dreams in and then begin a discussion with AI to see where it goes.
–
–
Where I got to last night was a question.
“How do I tie this dreaming project off?”
“How do I park it and move on?”
The obvious answer is to change the experiment. The easiest way to do this is to stop putting dreams up online and to see if that changes the dreaming…
If I close the blog I can concentrate on looking into the set of dreams and see…
By changing one variable I might alter my dreaming significantly….
Dreamyoga 