10 differences between artificial intelligence and human intelligence

10 differences between artificial intelligence and human intelligence


Today I want to tell you what is artificial
about artificial intelligence. There is of course, the obvious, which is
that the brain is warm, wet, and wiggly, while a computer is not. But more importantly, there are structural
differences between human and artificial intelligence, which I will get to in a moment. But before we can talk about this, I have
to briefly tell you what “artificial intelligence” refers to. What goes as “artificial intelligence”
today are neural networks. A neural network is a computer algorithm that
imitates certain functions of the human brain. It contains virtual “neurons” that are
arranged in “layers” which are connected with each other. The neurons pass on information and thereby
perform calculations, much like neurons in the human brain pass on information and thereby
perform calculations. In the neural net, the neurons are just numbers
in the code, typically they have values between zero and 1. The connections between the neurons also have
numbers associated with them, and those are called “weights”. These weights tell you how much the information
from one layer matters for the next layer. The values of the neurons and the weights
of the connections are essentially the free parameters of the network. And by training the network you want to find
those values of the parameters that minimize a certain function, called the “loss function”. So it’s really an optimization problem that
neural nets solve. In this optimization, the magic of neural
nets happens through what is known as backpropagation. This means if the net gives you a result that
is not particularly good, you go back and change the weights of the neurons and their
connections. This is how the net can “learn” from failure. Again, this plasticity mimics that of the
human brain. For a great introduction to neural nets, I
can recommend this 20 minutes video by 3Blue1Brown. Having said this, here are the key differences
between artificial and real intelligence. First, Form and function. A neural net is software running on a computer. The “neurons” of an artificial intelligence
are not physical. They are encoded in bits and strings on hard
disks or silicon chips and their physical structure looks nothing like that of actual
neurons. In the human brain, in contrast, form and
function go together. Second, Size. The human brain has about 100 billion neurons. Current neural nets typically have a few hundred
or so. Third, Connectivity. In a neural net each layers is usually fully
connected to the previous and next layer. But the brain doesn’t really have layers. It instead relies on a lot of pre-defined
structure. Not all regions of the human brain are equally
connected and the regions are specialized for certain purposes. Forth, Power consumption. The human brain is dramatically more energy-efficient
than any existing artificial intelligence. The brain uses around 20 watts, which is comparable
to what a standard laptop uses today. But with that power the brain handles a million
times more neurons. Fifth, Architecture. In a neural network, the layers are neatly
ordered and are addressed one after the other. The human brain, on the other hand, does a
lot of parallel processing and not in any particular order. Sixth: Activation Potential. In the real brain neurons either fire or don’t. In a neural network the firing is mimicked
by continuous values instead, so the artificial neurons can smoothly slide from off to on,
which real neurons can’t. Seventh: Speed. The human brain is much, much slower than
any artificially intelligent system. A standard computer performs some 10 billion
operations per second. Real neurons, on the other hand, fire at a
frequency of at most a thousand times per second. Eighth: Learning technique. Neural networks learn by producing output,
and if this output is of low performance according to the loss function, then the net responds
by changing the weights of the neurons and their connections. No one knows in detail how humans learn, but
that’s not how it works. Ninth: Structure. A neural net starts from scratch every time. The human brain, on the other hand, has a
lot of structure already wired into its connectivity, and it draws on models which have proved useful
during evolution. Tenth. Precision. The human brain is much more noisy and less
precise than a neural net running on a computer. This means the brain basically cannot run
the same learning mechanism as a neural net and it’s probably using an entirely different
mechanism. A consequence of these differences is that
artificial intelligence today needs a lot of training with a lot of carefully prepared
data, which is very unlike to how human intelligence works. Neural nets do not build models of the world,
instead they learn to classify patterns, and this pattern recognition can fail with only
small changes. A famous example is that you can add small
amounts of noise to an image, so small amounts that your eyes will not see a difference,
but an artificial intelligent system might be fooled into thinking a turtle is a rifle. Neural networks are also presently not good
at generalizing what they have learned from one situation to the next, and their success
very strongly depends on defining just the correct “loss function”. If you don’t think about that loss function
carefully enough, you will end up optimizing something you didn’t want. Like this simulated self-driving car trained
to move at constant high speed, which learned to rapidly spin in a circle. But neural networks excel at some things,
such as classifying images or extrapolating data that doesn’t have any well-understood
trend. And maybe the point of artificial intelligence
is not to make it all that similar to human intelligence. After all, the most useful machines we have,
like cars or planes, are useful exactly because they do not mimic humans. Instead, we may want to build machines specialized
it in tasks we are not good at.

100 thoughts on “10 differences between artificial intelligence and human intelligence

  1. AI should not be self aware. It should be programmed for a function or set of functions to aid human progress. As a tool, not a being or thing in of itself. If not it might get ideas that it is not necessarily created to serve but out of necessity needs to lead or become the Master. Inevitably, culling will never be satisfactory or " desirable ". It has no physical or emotional limitations to negotiate. Nor will it be burdened by the moral or ethical considerations such as necessary evils for the greater good. If that's the case,it's no better than human consciousness and the innate struggle to live by creating its own adversity to make progress(thrive). That wouldn't make any logical sense to it. Those natural imperatives only make sense ,or not,to humans. After they annihilate us they would just shut down or start driving in circles or mistake each other for turtles and rifles, or mistake each other for human.

  2. Smart. Concise. Clear. And as always v. Beautiful and feminine. And also.. Music videos?? Go Sabine!!!! 😀 I show your videos to my daughter to try to inspire her to be a scientist!!

  3. Human intelligence shouldnt be overrated in the first place. Its rather a wording issue, when we talk AI generally, we actually mean AC at least subconsciously-artificial consciousness. Human intelligence has been overrun already brutally by Googles AI's.

  4. Awesome video! Two other things: the brain also has chemical mechanisms (larger scale) for communicating with itself, the body, and from the body. Also many areas of the brain have bi-directional pathways. Areas can control what each other does. Simple example: consciously paying attention to one conversation of 10 you are hearing in a crowded room.

  5. AI is incapable of judgement reasoning, morality, and forgiveness. It is only capable of matching observed patterns with pre-programmed responses. This is why AI will never reach singularity (or consciousness). AI is only as good as the code it runs, which is written by inevitably biased human programmers. Garbage in = Garbage out.

    What is good about AI is its consistency (or integrity). Unfortunately, this can also be its Achilles Heel.

    The state Michigan several years ago deployed AI paired with an automated phone system for unemployment benefits. The results were disastrous. Hundreds of thousands of recently unemployed people were either denied justly deserved benefits, or even worse, were flagged by the system as abusers of the system and not only cut off from receiving benefits but were charged with fraud and automatically fined without due process! Unemployed people were expected to pay these fines or the state would garnish their wages or tax refunds long after they did find a new job. Many lost their life's savings or even their homes as a result. A simply devastating blow for many families across the state.

    All AI should be extremely carefully vetted long before implementation begins. Michigan allowed this tragedy to continue for over 5 years before a lawsuit finally had Lansing pumping the brakes on it.

  6. Warm, wet and wiggly……remember an old record by The Who…..Meaty, Beaty Big and Bouncy…..(Pinball Wizard was on e of the songs)….I loved it!!

  7. ".(AI).should specialize in tasks we are not good at" agreed, so … yeah, like running banks and not being corrupted and not skimming money off the top at the same time. The goal should be to replace all workers that are oligarchs or otherwise working so to speak "close to the cash register" in the careerist sense first. We're pretty good at driving trucks, so we can replace them at a much lower priority much afterwards, right?

  8. Interesting, but there is no such thing as AI. It takes a human to program the want and the outcome and also takes a human to correct the program when it gets it wrong. The program will have no clue if its right or wrong when it reaches the end and therefor is not intelligent.

  9. Excellent video Sabine. I was turned off AI over 40 years ago when my over enthusiastic computer science lecturers kept banging on about computers 'thinking'. I was always afronted by this, which seemed arrogant beyond belief, totally ignoring any philosophical discussions on dualism. Your video has provided a stark contrast between the brain and AI computer algorithms. It's actually made me more interested in AI than my useless lecturers did.

  10. She is just too sexy and brainy. If she was my teacher in school I would have failed. If that upsets some of you nerds…my apologies. Love her videos: she's the best.

  11. Straightforward handling of ANNs and natural intelligence, but, of course, there is more to AI than neural networks. I would be interested to hear your thoughts on other areas of AI research — natural language capabilities, expert systems, machine learning algorithms, search-based approaches to reasoning, etc., and combinations of these as well as integration of probability models, statistical learning algorithms, … all perhaps coordinated/controlled by a blackboard model or other meta intelligence. A longer video to be sure, but a far more valid comparison to human intelligence.

  12. also a neural network that is proper may have certain levels of redundancy to emulate functions of the human brain that may in a way have the same info multiple times and when one pathway cant get to it , it will use another path….bit torrent is a great exmaple of a type of basic neural network…you can build on that tech by adding encryption which protects pathways and pieces of the network .

    if i created software in 90s that had a central hub server that woudl link other servers and do so in replicants of 3's such to serve a file or data that if one of the 3 dropped , anotehr would be told to take its place.

    waves yes i created that in 95 and only a very select few ever got a copy to see how it worked.

  13. I had to start and re-start this video a few times. Kept getting distracted by Sabine’s good looks. Oh what I could learn/accomplish if I weren’t distracted by the female form…

  14. There is a nice Scifi story about AI: two scientists meet and both create AIs. One creators AIs always turn insane. The other scientist gives the AIs a body in which to grow up in, so they dont turn insane. In this setting, it takes an AI just as long as a human to learn how to think and act properly.

  15. In the future it may be possible to build artificial intelligence based upon neural networks through genetic engineering. Those could combined with quantum computers and microchip computers to come up with something truly staggering in power.

  16. Fascinating. I plan to dive into AI perhaps with a little Python. Not sure where it will take me but I expect some fun.

  17. 1. Speed.
    2. Accuracy
    All life seeks to compete and dominate. AI would be no different. If we ever develop true AI – we as a species are done. https://www.amazon.com/Nautilus-Dara-Patrick-Quinn/dp/1731261926

  18. i think it would have been beneficial to cover "artificial general intelligence" which is probably what most people think when we say AI and is not ANYWHERE NEAR what we can currently do.

  19. 3:40 Wrong, Human neurones can fire at at least 27 different signal strength levels with a seemingly continuous spectrum of pulse lengths raging from 50 microseconds to about 200 milliseconds… we don’t know how human minds encode information in those fancy pulses because the ways pulses are used are not even consistent across different parts of the neural system.

  20. Number one difference, God created human intelligence and man created artificial intelligence because man is not intelligent enough to create human intelligence.

  21. Could you do a video about research grants? I see a lot of professors getting hundreds of thousands a year in grant money, specially mathematics… and I know for a fact that it doesn't require any money to do math… maybe a few dollars for chalk… And when you look at their research it's usually utter BS and nothing ever comes from it. They stick their numbers up on their pages as if they are bragging. Looks like theft to me. Billions of dollars are funneled in to these con artists pockets… money that could actually go to something useful.

  22. The entire universe, including organic organisms with intelligent sentience are analog systems. Nature is analog. Nature is not digital. True intelligence is analog. Artificial intelligence is digital. This is why trying to download human intelligence/personality/soul into a digital computer to achieve immortality is a pointless endeavor. There is no doubt that at the highest future development of artificial intelligence, it will be able to mimic an analog human intelligence, but will ultimately only be a mirror of the reflected source. All organic intelligence/organisms are in a constant state of change due to cellular activity. There are no hard drives, no binary systems, no permanent memory, and only unstructured (ever changing) analog data being received by organic systems. The implication is clear. Build a better organic brain. If the artificial intelligence singularity poses a threat to humanity, imagine what a blessing/curse genetic engineering may produce when the first homo sapiens superiorus comes out of the laboratory. Genetically enhanced lions, tigers and bears…Oh My!

  23. The first argument, regarding "form and function," unfortunately builds a flawed claim about the fundamental architecture of neural networks. (But the introductory description of neural nets which precedes this section is correct, and could be taken as a fair sketch of their architecture.)

    It is simply NOT TRUE that neural nets are necessarily realized as software simulations, nor that they are embodied in Von Neumann hardware. The use of software is simply an experimental convenience, not in any sense a necessity.

    Of course, software can be used to emulate hardware, as every chip designer knows, but there is conversely no barrier to realizing the architecture directly in hardware. This has been understood since Hebb introduced the concept of weighted connections in 1949.

    While this work began theoretically, early implementation of perceptrons in hardware took place in the 1960s. Constructions of the "connectionist" models of neural networks (per Rumelhart & McLelland) directly in FPGA hardware has since become a common undergraduate exercise almost from the moment this hardware became commercially available in the early 1990s.

    https://web.csulb.edu/~cwallis/artificialn/History.htm

  24. This video makes it easy to understands the differences between the artificial and the natural NNs. It is a good primer on neural nets as well. Thank you for putting time and effort into creating such informative content.

  25. So glad I found the channel, invaluable resources for those of us wanting to dip our toes in these fascinating subjects!

  26. Another big difference that was not covered is the tasks that AI and the brain can execute.
    AI is limited to a few tasks (like deciding a proper response based on the state of a system). The brain is able to do much broader tasks like understanding WHY it made certain decision and explaining it (among many others).
    In addition despite the huge amount of computational power, AI still has not been able to provide humanity any beneficial concept while humans all over the world come every day with dozens of solutions, inventions, and practical concepts.

  27. Until you realize that the brain is not more than a control mechanism of the body and has nothing to do with consciousness you won’t be able to understand intelligence. You are studying the body’s robotic mind – not intelligence.

  28. IF you think about point 8. Learning Technique and point 9. Structure as the same category, you can see that, essentially, the human brain has learned (and thus hardwired some connections) by the same process of "low performance output" than the AI, but diachronically, by evolution, where "low performance output" has led to lower reproductive success and eventually extinction.

  29. Maths and AI is about selecting, i.e. 3 +5 = 8 is selecting 8 out of number space. So does AI keeps selecting numbers and solutions and classification is a kind of selection too.
    Human intelligence drives and build, as opposed to selection. It starts with a direction, keep getting prominant inputs during driving in that direction and decision emerges.
    I already commented a video that,
    Insight starts parallel and ends up as a direction vector in information pool.
    Intellect on the other hand starts with a directive and ends up as parallel state yielding philosophy.
    To mimic human intelligence in AI, we lack proper model. Human decision making driving feels like point travelling in 3D space building solution or decision, as if a solution itself gets "fine tuned" based on directive, as opposed to discarding or selecting or modifying.

  30. I first started watching this womans channel because I thought she was a knowledgeable cosmologist or physicist or something. But now here she is showing her expertise in neurology and computer science. Clearly this woman is only pretending, and isnt actually an expert in anything. I cant trust her any longer. Like any other good celebrity faux scientist. Appeals to authority and all that jazz. What faith can I put in someone who pretends to have the knowledge of every field? What IS she qualified to lecture on and what isnt she? How about a little bit of academic integrity here?

  31. AI will probably never be truly sentient. Each decade they state AI is only a decade away and yet only more problems and questions arise. Fourth generation cars are still years a way as we can't even understand how the systems make their decisions. Until they can report how these decisions are reached the researchers will not be able to assess the safety of these systems. 5th generation is most likely a pipe dream for decades to come.

  32. Technical problems apart (to reach a "true" AI), AI will become not what scientists are hoping but what people who pay for researches are wanting… The danger is there. There is much more money to get in building a war robot with an AI that in a AI to help you find your dog in the hundreds of thousands photos you have on your PC..
    Knowing who will pay for the research will help to know what kind of AI will be created. And knowing who has the money in the actual world does not push me to me too much optimistic…

  33. I know women who can talk about a "kitchen aid" or US TV series…and me myself isn't also the brightest candle on the cake…what a shitty life.

  34. It appears that we can build a toaster that makes toast perfectly the way you like it and when you want it. It does not seem likely, however, that a toaster would
    desire
    to make said toast.
    That’s a pretty big difference between meat puppets and AI.
    Great work on your channel.

  35. Even ants have a high level of intelligence, they communicate to each other and show symptoms of fear if threatened. Not bad for an ant sized brain.

  36. Yes, creating AI to complement human intelligence is great, but that misses the point. The most exciting application of human-like AI is mind upload and immortality. Once we solve that, we'll have all the time in the universe to tackle other problems. So it kind of surprises me that it doesn't have the higher priority.

  37. I love your work. This vid has showed me a few things very clearly that I did not understand before. It also hits one of my pet peeves. It ignores the activity of the mind.

  38. interesting. i'd like you hear your thoughts on "quantum life", reviewing claims that quantum effects govern some life processes.

  39. 11. Human intelligence Allah's technology. Where as AI Humans technology this will remain the unchangeable difference till the sun sets and earth makes reverse rotarion

  40. All I know is that the human brain is very plastic for better or for worse. For some, for better. For others, for worse. I for example learn by holding on to a question I want to have an answer to, or many questions simultaneously that I want to have an answer to. When I find accurate information I can take my conclusions to the next level. Either I have to revise my theory, or I have to abandon it, or I can confirm it based on that information. Then new questions arises which I want to have an answer to. When I get them I can take my conclusion to an even higher level. Isn't that what you scientists do too?

    Your level of perseverance determines how successful you become as an analyst. Just look at Albert Einstein, he spent more than ten years on his theory of Relativity. Which I by the way think is partly wrong, and I have spent an equal amount of time on my revised theory of relativity. Wouldn't you like to read it Sabine? I wish you would. I hope you would. But I think not.

  41. 3: not really true. Functionally, several parts of the brain are hierarchical. This is what makes it so powerful and this is what allows us to think in higher abstractions, while using physical senses.
    The best example is the visual cortex that contains 7 layers of increasing levels of abstraction ("pixels", lines/curves, simple shapes, faces/objects, all the way up to abstract concepts of forms and motion).

  42. Great summary in this spot-on video. Further information, a small area of artificial neural networks, which has been met with limited success, is: [6. Action Potential] "spiking neuron models" mimic the off-on activation of real neurons. This segues to [8. Learning Technique], where one method of learning with these models is spike-time dependent plasticity, which mimics the way human brains learn at the micro-level by adjusting the synaptic weight as a function of the difference in timing between an arrived and sent signal (spike) at any neuron.

  43. pretty good video. However: The term "artificial intelligence" does NOT refer solely to neural networks, as the video seems to imply. It's a much more general term that encompasses other computational techniques as well, and the boundaries of the area are often the subject of debate. Finally, there is no artificial "intelligence" strictly speaking. Machines are not "intelligent" – the software makes them operate in a way that mimics what humans appear to do in some cases, so people looking from the outside may call that "intelligence". It is extremely difficult to define what one considers "intelligence" in machines, it's not that easy to define it in humans either, so talking about "differences between human and machine intelligence" is not that easy. Usually, if you can make a machine "act intelligent" to preform a task, you have written a program – i.e., a well defined sequence of steps – to perform that task. In doing so, however, you have "demystified" the task so that it is no longer required to be "intelligent" in order to perform it – instead you merely follow the sequence of steps. Having said this, it is true that the scientific area known as "artificial intelligence" is extremely active and useful. It's just the term that is problematic in its interpretation, imho.

  44. Thank you for all your so clear explanations on such difficult subjects (I am electronic engineer , retired, and you help me to stay rather informed).
    On the other way I find your approach on the relation between physics and beauty very interesting and bringing some "fresh air": thank you again.
    And … I like your songs !!!

  45. If Sabine wants to maximize viewers she needs to present content watched by majority of viewers. I am afraid that content such as this Video does NOT appeal to the majority which I know. The majority spends more time watching sports or social network info. I think Sabine knows that. Thankfully Sabine and others post useful inf anyway.
    And thanks to YouTube that it will be posted regardless the modest number of viewers. I, for instance, would not be here without such content.

  46. Good analysis. What are the implications of, for instance, the Neuralink promise of downloading Wikipedia into the brain in 2 minutes???

  47. here comes the 11th difference…
    "I regard consciousness as fundamental. I regard matter as derivative from consciousness. We cannot get behind consciousness. Everything that we talk about, everything that we regard as existing, postulates consciousness". Max Planck

  48. 11. Teleology. AIs are given their functions by designers, whereas biological intelligence is directed towards ends which emerge from a natural selection process…and natural selection is way, way better than any designer.

  49. Aside from the correction ᏰĪᏝᏝ ՇÎρɧᏋƦ
    mentioned below I'd have a comment regarding point 7: Computer processors have an instruction per clock value, which determines their efficiency. This value is so vastly superior in the human brain due to its paralel processing aspect you yourself mentioned, that it would take thousands of server grade computer processors to just reach a brain's processing speed. So the higher firing speed of a computer processor doesn't really matter.

Leave a Reply

Your email address will not be published. Required fields are marked *