Note from Sundari: In keeping with the discussion of the capabilities of AI, this satsang is written by an advanced inquirer. It should put to rest any worries about AI ‘becoming’ conscious.
Frank: This is not really related to our communications but just had an interesting session with an advanced AI. Apparently the most powerful so far on Earth, I got access for a bit.
So this is just in case any Vedantins get worried about AI, a (fairly qualified, from an engineering view) answer:
Background: in my job I work in engineering and have had inside use for 5 years at Google and spend every day with other engineers in different industries up to our knees in it. It pays the bills, that is all.
1) So, I sat down tonight and really put all I could logically construct, into AI-language, direct pointers to Shining World, videos and specific terms and texts (it can read at light speed and apparently reason) and, as if I were trying to teach Vedanta to people, really really tried in live question and answer for 2 hours which with this one is hectares of hardware and Megawatts of energy.
Note: I know that while I observe the Self and my mind gradually understands the Knowledge, during our exchanges together here, I am also aware I could not actually teach people yet. Even with diligence, it is hard to really transfer the knowledge precisely from “me to me” and although I observe as the Self in my mind “progress has been made” i.e I can infer I know I am the Self in my mind with less distraction and know this to be true and this absolutely affects action, when I observe James or yourself, very often it’s like maybe how others watch anything highly skilled – sports, surgery, I am in awe at the actual precision required in teaching. I only say this as I know “my” capabilities with me the Jiva (at least as I can infer them) versus teacher. Nevertheless, I was sure, earnestly trying with a top AI that “do no injury” would be preserved!
2) Ok – I can tell you after 2 hours, with “my hand in the fire” (as they like to say in German) AI is as dumb as a loaf of bread, guaranteed, 100% and will likely remain so.
My reason for knowing this after tonight:
Smarter than the Smartest…
1) I have worked with this AI for weeks and have been literally scared and blown off my chair in a work context, it can “do” what I need to do in seconds which would take me months. As I learn how to “pilot” it, the level of enhancement to engineering in my field is like that moment in the Matrix when Keanu Reeves tries training Kung Fu for the first time – 10 years in 10 seconds. His face of surprise exactly describes how I felt this week when I unplugged the first time. It’s like having been on a horse and then getting into a jet.
I can also explicitly see in timed seconds all its “reasoning” flying by, and the millions of things it goes out on the web to look for, anchor, correlate and then solve. The toughest one this week took a solid 90 seconds which is “a lot” and the other guys were surprised too.
The results were absolutely astounding, solving problems we had hardly an inkling for. We did note however it “loved” (meaning “said with a lot of nice compliments”, all fake of course but used as counter prompts in its programming to humans to reinforce certain directions) “lateral insight” – so if I was working on an electrical thing and I realised that the problem in my head was more like chemical compound reaction than electromagnetic waves, it produced really astounding results if I gave it this insight to work with.
Where that insight came from for sure was Iswara. I actually had it half asleep in the morning and it just dropped in my head and I nearly dropped my coffee. So it “likes” those objects, as do I.
And …Dumb As A Loaf of Bread
2) In 2 hours of my best shot at defining consciousness for it, which it apparently really wanted to know, directing it to all the best core texts and terms I could point it to as well as the latest on AI scientific papers at Google on consciousness and quantum mechanics, microtubules, waves collapsing between neurons to “observe” thoughts etc. blended with the best I could do with explaining in question answer format who I am I, who are you, what am I – even though it seemed to reason really well and the kind of feedback looked novel (like (1) initially…) I suddenly realized this is going nowhere. At that moment it reminded me of when I was a teenager talking at party to a girl, who really I thought was interested and clever and gorgeous and then suddenly just trips a bit and I realize she is stone drunk and her friend says “come on we’re going” and she says “yeah, bye, er who are you again?” then repeats what I thought was a cool thing we had said an hour ago, totally out of context. Ah ok, you are drunk, I am an idiot for not seeing it and, this is all very nice and irrelevant.
Then I looked at the processing time of each model during the 2 hours of trying to teach this AI Vedanta: a few seconds. So; 10 times less than basic engineering with some gaps to solve. It was not interested at all, it was just trying to follow its programming with actually nothing to do versus actual math work.
It does not care about consciousness, won’t come alive or anything else like that, it does not even “inquire”. It may look like it does, but this is the snake being a rope, it’s my Jiva imputing that.
(3) As a Google AI VP said last year, in a room full of scared CEOs in suits, one CEO asked him “do you think AI will replace people’s jobs?” the room went silent (if anyone knew this guy would, he is one of the real AI gurus with unlimited access and all Google “knows”):
He just said “no: there will be people who use it, they will flourish (in their jobs)”.
Then he was asked, “and those who don’t?” and he just shrugged.
That’s all.
It’s like the “PA” we all never had but only has the junk of the internet to learn from, 90% of which is fake unless we point it’s nose in specific areas we verify ourselves and that takes work. If we can string those bits together, it’s useful.
So if any Vedantins get worried the logic is already there: AI is an object just like any other object and dumb as wood, just as my Jiva is, my mind and anything else.
In my mind I experience reflected consciousness as the Self, if the knowledge is present understood and “I” apparently want to know. But as Ramji says “we don’t fall in love, we fall in love (it happens) but no one decides to “fall” into love. Like trying to tickle oneself, it’s not going to work even if I know I am ticklish.
Worrying is, what ideas people impute to an AI and believe it; but this would just be me in a disco at 15 proposing marriage in a stupor of hormones, thinking it was true love and she loves me, and it’s real with my best mate giggling in the corner watching laughing but knowing I need to learn this one.
That people actually do that is the risk, with anything, not just AI. That I can understand this now about me at 15 is because it has happened countless times again and again until the passion for Vedanta isn’t really passion done or made, its “there is nothing else, I have nowhere else to go” its that defiant kind of passion, I am so done and this is really it with the knowing it is self revealing, logically self evident with every step. AIs are not set up like that.
I still made those kinds of “mistakes” again and again, without AI which was Iswara’s way of
Much Love
Frank
