ENTP Teaches Ai to See

Alexei Efros is a computer scientist (and alleged ENTP) who was loosing his sight while teaching Ai to see. Really fascinating stuff helping us understand the observer functions better.
Why Computer Vision Is a Hard Problem for Ai: • Why Computer Vision Is...
For more about cognitive functions head to www.ObjectiveP...

[@jks234]

Fascinating. It's fascinating how AI is revealing just how complicated our processing circuitry is. As you point out here, we don’t actually directly see much. Rather, we remember. And thus, vision is perhaps 10% a sensory problem, and 90% a memory and pattern matching problem. Humans have some fascinating cognitive interactions going on. And MBTI is a great way to parse it and map it out.

[@RealnessEternal]

As an ENTP the fact that labelling objects doesn't lead to robust intelligence is perversely something that I'm very happy about ("See! The labels aren't the reality guys!) I'm going to speculate on the evolutionary history of the jungian functions with little evidence and some pattern recognition so consider my opinions just pure speculation. I suspect that the extraverted sensation is the oldest function since you need an integrated experience of different sense organs in order to benefit from having sense organs at all (no point having eyes without a brain). Then introverted intuition would be the next oldest function since animals that have brains can learn to interact with their environment in different ways including interacting with other conscious agents (cambrian explosion). This sets of an arms race between predator and pray in that understanding that there are other conscious agents vastly improves hunting/avoiding being eaten. Brains that have a hippocampus (like a mouse) can generate an internal model of the world that can be interacted with (i.e. they have object permanence). This would correspond to both introverted sensation (giving a simulation of physical reality) and extraverted intuition (able to explore possibilities in this simulation). This would let a mouse imagine solutions to navigating a maze for instance. It seems that every animal needs to have a capacity for sexual desire in order to mate so some kind of feeling/desire is going to be conserved for every animal. I suspect animals that have to care for young need something like extraverted feeling and animals that need to function in a pack or shoal need something like extraverted thinking to process social hierarchy. So are cats ips and dogs eps? When OPS for humans is done the next step is Pet OPS:)

[@AustinOllar]

Your recent vids are great Dave. Gonna try not to crash into a bunch of kids at the bus stop 😂

[@AITheWise]

The video discusses how AI systems are learning to see and understand the world, moving from supervised learning to self-supervised approaches that mimic how humans perceive their environment. Key moments: 00:00 The video discusses the intersection of AI and Ops, focusing on how AI is used to understand vision. It highlights the importance of connecting sensory input with past experiences for AI to 'see'. -The speaker shares a personal story of how his vision loss led him to work on AI and vision, emphasizing the role of memory in compensating for poor vision. -The evolution of computer vision is explored, revealing the shift from focusing solely on algorithms to realizing the significance of connecting sensory input with past visual data. 03:56 Computer vision has two paradigms: supervised learning with labeled images and self-supervised learning without linguistic supervision, mimicking how animals understand the world. AI's potential benefits for humanity outweigh doomsday fears, with researchers exploring AI to understand the human brain and relationships. -Challenges in programming AI to react to real-world scenarios like a plastic bag crossing the street, highlighting the complexity of decision-making in autonomous vehicles. -Comparison between supervised learning using labeled images and self-supervised learning without linguistic supervision in computer vision research. -Researchers, including tech leaders like Bill Gates and Elon Musk, studying AI to gain insights into the human brain and relationships, emphasizing the interdisciplinary nature of AI research. Generated by sider.ai

[@coolcool2901]

Intelligence is basically Pattern Recognition and Knowledge Representation, which power Fluid Intelligence and Crystallized Intelligence.

[@brucebruno842]

Yes, AI intelligence. AIs may be able to retain knowledge and build a greater and greater understanding of that knowledge, but wisdom, how to properly apply the understanding of that knowledge; which comes from real experiences and consciousness it doesn't have. Feelings are also information expressed in the negative and positive; which is the last piece it's missing intelligence wise.

[@coolcool2901]

@@brucebruno842 All these things you are talking about are powered by pattern recognition and knowledge representation.

[@statisticalresearchmethods]

Machine learning is a form of computational statistics. Data scientists are teaching AI to see patterns; therefore, anyone interested in OPS should see computational statistics as promising.

[@chronomirageYT]

I feel like current AI is more like Se and Si, because the know info is BIG data, not intuitive, that's why they not as good for now. Maybe when it reaches Ne-Si or Se-Ni it's gonna be good.

[@gustavowantony2476]

May i ask what you meant there? Because if Ni and Ne are not contained in the raw data, then where is it (or would be) contained? It's just when we talk about it like that, we make it seem as if intuition is this fairy dust thing that doesn't physically exist, which is not correct. The reason i ask is because i've largely studied neural networks for a long time now and it's interesting to see other people's views on the matter, but also, as an enthusiast of the OP Code i do think it has a lot of similarities but also fundamental differences.

[@chronomirageYT]

@@gustavowantony2476 I think in OPS intuition means pattern, As human we sometimes forget the data and only remember the pattern. I see my tesla's has something called vector space, maybe it's like that? But I'm not qualified to have my opinion taken seriously, Idk the technically detail of AI and comptur vision.

[@gustavowantony2476]

@@chronomirageYT Oh, I think I see what you mean now. So, in information theory there's a concept known as a bit of information, which is a literal physical thing that conveys some information about something. Therefore we conclude, if intuition is information (which it is), it must be physically stored in the brain somewhere, even if it's information about the patterns in other information, makes sense? It's like a set of data encapsulating relationships between other sets of data, it's still all data. Now about the vector space in the tesla, I don't really know what that is in that context, but a vector space generally os just a finite list of numbers that represent something (generally coordinates, and can even have more than 3 in a hyper-dimensional coordinate system. Despite our brain not being to able to understand more than 3 dimensions, math still works there). Maybe it's used on cars just like in the video, to numerically comprehend distances, directions, and so on

[@RealnessEternal]

I think of chatgpt as having superhuman Se in that it has "read" far more language then a human would even be capable of. I think of it as having limited Ni since it has to abstract to a limited level in that it can't really make intuitive jumps that a smart human would. I don't think of it as having a world simulation or internal experience so I don't think it has any Si or Ne. This is why it can often make basic mistakes that a human wouldn't make.

[@gustavowantony2476]

@@RealnessEternal If we really had to explain artificial intelligence on a lower level, you'd be somewhat correct. The thing is, neural networks don't actually *know* anything, i've been where you are and i can assure you, when you know little about how it works it does seem like superpowers or magic, but in reality it's pure and utter math. What an AI actually is is a very big and fancy mathematical function that has parameters (a bunch of numbers it uses to calculate), then you give this network some inputs (say, the pixels on an image) and based on each pixel value it calculates a final output (say, whether or not it thinks it is a cat). But we humans can't possibly understand how we would choose those parameters right? That's why we don't, we just start them randomly which just makes the network produce very bad results, and training is responsible for tweaking each of the hundreds, thousands, millions of parameters that will best work for all the possible images of cats/not cats in the training set. The AI will just learn the best numbers to predict from the data it was trained on, there's no "abstraction", there's no "internal experience", there's no "simulation" at all. It's just numbers changing to do what we want them to

[@ethanotene2993]

Uncle Dave!!!!

[@ElectraQvQ]

I love breaking AI as an entp :>

[@moneyvsfinance]

A.I. is simply a computer program, that, instead of given instructions, is given a set of actions it can take that will be scored on the desired outcome. Each score affects the probability that action will be taken in the next simulation, obviously making the 1st simulation completely random (unless otherwise specified by the developer). So, A.I. can easily be a tool, or a weapon. So the question of whether or not A.I. will be good or bad is not about the A.I. in of itself. It is about who is behind what A.I. Unfortunately, our world is run by a currency monopoly called the Federal Reserve. Inherently evil in it's nature, and whose owners have been publicly known to be behind the World Wars, and the statist regimes of foreign nations that have started them, will, by all rational calculation of the odds, be behind the most powerful A.I. Also, the hypothesis of Darwinian Evolution is stupid. And you believe it without knowing any evidence to support it. No some scientist saying "the fossil record" is not evidence. They really have nothing but conjecture. You should actually question it, not just blindly believe it, which you are obviously doing.

[@OptimusCrime4444]

As an xNTP i feel called out. All I wanted to do is destroy the world just a little bit so I can connect to my SF. And the world cannot even let me do that in piece?

[@Magani79]

peace - grammar police

[@lupoi14]

Piece as in Si

[@actazrath5878]

Sing bud, sing!

[@5idi]

0:55 No, it makes zero sense. Bring Shan in.

[@AdamMiceli]

Why does it make 0 sense? Dr Joe Dispensa and Dr Bruce Lipton have done a lot of good science showing that mental states can determine gene-expression. People have been documented to change their biology using primarily visualization practices. It might make 0 sense because you aren’t really studying it.

[@AdamMiceli]

And when I say “‘mental state.” I’m really talking about a psycho-physiological state. Emotions and thoughts aren’t only located in the brain. They affect the entire body which affects the environment that your cells are living in. This determines how your cells choose to express their DNA. Learn more about it in the authors works I suggested above.

[@orangeziggy348]

Isnt just Si the observer with past memory? Then how does Se+Ni see the world- wouldnt it need to see with the present (Se) instead of seeing with the past(Si)?

[@drwizarrd4421]

There's no such thing as Si = past. Not here anywhere, that is some 16p crap. Si is pretty much just organized sensory

[@filipposa212]

AI is the image of the beast - it will bring in total control of humans. However, this stuff is fascinating and really interesting to learn

[@accountnew7030]

huh...? u need to touch some grass occasionally ya know. ai is nothing but a bunch of mathematical functions that calculate probability. it has nothing to do with global control or conquering the world. if u think that the current ai has any sort of actual intelligence then you don't understand what ai really is. we very far from achieving something like AGI and it'll remain this way for quite some time unless there's an insane breakthrough in the field but right now ai development kinda hit the plateau after the initial ai hype wave. and if by calling it an image of the beast u mean it in a conspiracy/religious sense then you just need help buddy. I hope u get better 🙏🏻

[@filipposa212]

@@accountnew7030 it is the stuff that will enslave humanity. Just take a look at China

[@RealnessEternal]

People are already very influenced by the AI that generates recommendations and the AI is in effect programmed to try and get people addicted to social media. There is also the issue that AI is programmed to censor videos and comments so the politics of whose values are used to control what is allowed to be seen is hugely effecting society.

[@actazrath5878]

I think mbti is made to relate human and ai hands down.

[@Dolph-fe2ks]

So, are you saying, bridge TE/SE to Si/Ni.. Or, maybe vice/versa? 🤔

[@ADeeSHUPA]

why Te/Se

[@Dolph-fe2ks]

Because TE/SE *actually_is.*

[@audy2174]

I don't think Se/Ni is the "first" -- I think that's Dave's bias. Extroverted perceiving would be the first (Se is taking in the "Light", or the "Sensation" and Ne is taking in the "Shadow", or the "Nonsensation"), you cannot have one without the other, cannot have light without shadow and introverted perceiving (Si/Ni) would be what follows after -- the development of the "database" of all that one has sensed and predicted (Si) and what one has *not* sensed and predicted (Ni? let me know if this is a good description of Ni as it's my blindspot). Though I think the jury is still out on that because we are probably, as little swimmers in the womb, preloaded with some kind of instinctual memory that gets inherited as they've done experiments involving rats and their infants separated from birth from their parents who retained their parent's memories of environmental danger.

[@infensus1992]

You are perceiving the outer light as primary when the inner Light is source. The Moon reflects the Sun.

[@RealnessEternal]

I think oldest would relate to oldest in evolutionary time. The earliest animals with integration of sense organ input and motor output would need something like extraverted sensation. Animals that can recognise the difference between living and non-living animals and treat them as conscious agents need something like a primitive introverted intuition, Animals with memory and imagination such as mice need something like introverted sensation to create a model of the world and something like extraverted intuition to imagine changes in this internal model. Looking at it this way the jungian functions are ancient and conserved brain pathways.

[@audy2174]

@@infensus1992 Yes, the moon is reflecting the sun's light is my point. Both outer light (big bang) and inner light are Source. Look at dimensionality, Penrose diagrams, Infinity symbol, oroboros, etc.

[@audy2174]

@@RealnessEternal That's exactly how Jung was expressing he didn't want his theory interpreted in that manner. All functions exist in all, all at once. Even simple bacteria are more complex than we give them credit for, they absolutely have Ne and Si.

[@RealnessEternal]

@@audy2174 I was expecting criticism for speculating on jungian functions when we can never directly ask animals what their experience is. While I'm familiar with pantheism (that everything is a form of consciousness or awareness) I didn't know there was a jungian form of it. Something more for me to consider! Thank you!