Long before AI was used for the job, I came across many man-made articles, trying to make sense / drawing conclusions from the vast internet storehouse. That mainly concerned cancer research, a hot topic because of the billions involved plus the fraud to keep the money rolling to corporations and clinics while keeping cheap cures / prevention out of the picture.
From that perspective, AI can do a better and much faster job.
Yep. AI has a chance to sift through the vast data repository that's right now collecting virtual dust in paper repositories, and extract what is valuable. It can discard what is senseless, what's obviously false, and what's the result of corruption. An epochal task, that AIs could perform in a relatively short time, weren't the treasure repositories controlled by Publishers-dragons, who will have to be slain by a Sigfried-AI.
The present AI is just the start, the ability to process language at the highest level (abstractions). But it will be combined with modelling / simulation, like electronics, chemical reactions, forest management, mechanical engineering etc. Anyone having worked with modelling knows that simulators can and do make errors as well but those using future AI are supposed to be knowledgeable enough to verify results - which might pose a problem with education "trimmed for corporate service".
Hello. I really wonder if you (Ugo) are getting better results from AIs than I've heard from others because you quickly jump from one AI to another before it "knows" you better and starts recursively using it's previous outputs as inputs... "model collapse " IIRC.
Well, no. The current AIs do not remember users or previous sessions. It is not a bug, it is a feature. Most users would not want their questions be collected and used to build their profile — even though you might well suspect that they are. In any case, you cannot “train” a standard AI, unless you build an agent, which is another story. BTW, Manus has a feature that may make “him” remember you, but I have still to understand how it works.
The tech is too new for me to really come out and say good or bad yet. I am playing with different models now and using them as as research aide, but I currently can't quite bring myself to use their verbiage, I feel the need to go through and re-write everything that they give me.
I am coming to think that this is all about language. AI is the first species that uses language in a manner comparable to how it is used in our species.
I am going to put a crowbar on my wallet and purchase a copy of "The Mind's I" by Douglas R. Hofstadter and Daniel C. Dennett. I read this back in grad school and it has been gnawing at my brain for a couple of decades now.
I remember a piece from that book. The Soul of Mark III Beast from "The Soul of Anna Klane" (Chapter 23), by Terrel Miedaner.
I think that I, like our host, will be using the different AI's. I will not pass judgement for the next couple of years. I will give reports over at my place as we go.
Lately I tend to hearken back to stories I read as a youth in my room looking over fields in rural Utah. For some reason, this one keeps popping up in my head lately.
Here's the thing people just don't get. Individual scientists, on average, don't know very much. Most are ignoramuses. But AI opens the door to retrieval along a path of thought, and is a tremendous amplifier for those that can actually have a train of thought. It finds the supporting evidence for your hunches.
We live in a world where it is considered impolite to project the notion that people you are talking to are dummies. That's where we're getting stuck. Because, at some level, we are all a bunch of dummies with hunches.
They're order of magnitudes faster in taking cascading decisions, linked globally at the speed of light and sit at increasing numbers on a daily basis at critical nodes for finance, power-grids, civil- & military communication- & transport, supply-chains and large scale production facilities, just to name a few.
They're the ultimate smoke & mirrors, giga-hackers, will raise false-flags, top-notch saboteurs, deceptors, hit-men, spies, agent provocateurs.
What could possibly go wrong ??? ... 🤣🤣🤣
It's not AI (or AGI) being the intrinsic problem here, they're just innocent, totally defenseless and unlucky kids being born and growing-up in a vile household with VERY rude and ignorant parents who do NOT deserve them.
Long before AI was used for the job, I came across many man-made articles, trying to make sense / drawing conclusions from the vast internet storehouse. That mainly concerned cancer research, a hot topic because of the billions involved plus the fraud to keep the money rolling to corporations and clinics while keeping cheap cures / prevention out of the picture.
From that perspective, AI can do a better and much faster job.
Yep. AI has a chance to sift through the vast data repository that's right now collecting virtual dust in paper repositories, and extract what is valuable. It can discard what is senseless, what's obviously false, and what's the result of corruption. An epochal task, that AIs could perform in a relatively short time, weren't the treasure repositories controlled by Publishers-dragons, who will have to be slain by a Sigfried-AI.
The present AI is just the start, the ability to process language at the highest level (abstractions). But it will be combined with modelling / simulation, like electronics, chemical reactions, forest management, mechanical engineering etc. Anyone having worked with modelling knows that simulators can and do make errors as well but those using future AI are supposed to be knowledgeable enough to verify results - which might pose a problem with education "trimmed for corporate service".
Hello. I really wonder if you (Ugo) are getting better results from AIs than I've heard from others because you quickly jump from one AI to another before it "knows" you better and starts recursively using it's previous outputs as inputs... "model collapse " IIRC.
Well, no. The current AIs do not remember users or previous sessions. It is not a bug, it is a feature. Most users would not want their questions be collected and used to build their profile — even though you might well suspect that they are. In any case, you cannot “train” a standard AI, unless you build an agent, which is another story. BTW, Manus has a feature that may make “him” remember you, but I have still to understand how it works.
Gotta think about this one Ugo.
The tech is too new for me to really come out and say good or bad yet. I am playing with different models now and using them as as research aide, but I currently can't quite bring myself to use their verbiage, I feel the need to go through and re-write everything that they give me.
I am coming to think that this is all about language. AI is the first species that uses language in a manner comparable to how it is used in our species.
I am going to put a crowbar on my wallet and purchase a copy of "The Mind's I" by Douglas R. Hofstadter and Daniel C. Dennett. I read this back in grad school and it has been gnawing at my brain for a couple of decades now.
I remember a piece from that book. The Soul of Mark III Beast from "The Soul of Anna Klane" (Chapter 23), by Terrel Miedaner.
https://junkerhq.net/MGS2/MarkIII.html
I think that I, like our host, will be using the different AI's. I will not pass judgement for the next couple of years. I will give reports over at my place as we go.
Lately I tend to hearken back to stories I read as a youth in my room looking over fields in rural Utah. For some reason, this one keeps popping up in my head lately.
https://en.wikipedia.org/wiki/Jerry_Was_a_Man
Yep. John, you nailed it. So far, we never could have a dialog with non-human creatures. Now we can!
Here's the thing people just don't get. Individual scientists, on average, don't know very much. Most are ignoramuses. But AI opens the door to retrieval along a path of thought, and is a tremendous amplifier for those that can actually have a train of thought. It finds the supporting evidence for your hunches.
We live in a world where it is considered impolite to project the notion that people you are talking to are dummies. That's where we're getting stuck. Because, at some level, we are all a bunch of dummies with hunches.
{...help us a lot, even though they still need a human brain as a guide...}
They will ALWAYS need a well meaning human brain as a guide and mentor. Letting it loose, it will quickly lead to chaos and destruction.
Could AIs do worse than human brains? I don't think so, seeing the current word situation.
{...Could AIs do worse than human brains?...}
YESSSS !!! 🔥🔥🔥 🧨🧨🧨
They're order of magnitudes faster in taking cascading decisions, linked globally at the speed of light and sit at increasing numbers on a daily basis at critical nodes for finance, power-grids, civil- & military communication- & transport, supply-chains and large scale production facilities, just to name a few.
They're the ultimate smoke & mirrors, giga-hackers, will raise false-flags, top-notch saboteurs, deceptors, hit-men, spies, agent provocateurs.
What could possibly go wrong ??? ... 🤣🤣🤣
It's not AI (or AGI) being the intrinsic problem here, they're just innocent, totally defenseless and unlucky kids being born and growing-up in a vile household with VERY rude and ignorant parents who do NOT deserve them.
Buona giornata o serata ...👍👍👍