Check out Foreo at foreo.se/cy3q and get 30% off UFO 3. For the first 50 people, get a 10% additional discount using the code 10SIDEPRJ. Thank you FOREO for the sponsorship!
I was chuckling a bit when the example of whales came up as stupid because we are a prime example of stupid too, well, our eyes specifically. You see, pun not intended, our eyes evolved the wrong side around, that's the reason why we have a blind spot.
The main misconception of evolution is that all adaptations we have ever witnessed is the loss of dna not the gaining of dna. Along with the huge lack of transition species in the fossil record. That was Darwin’s biggest fear and it’s only made the idea of evolution worse. The cell example given only shows micro evolution and adaptation which is what I discussed earlier about dna being lost or shutting off for adaptation, it never adds dna. So those single cells have never been observed becoming multi-cell. Macro evolution has never been observed. Also when it comes to creationists, evolution works just fine biblically and structural evolution sinks up perfectly with a creator. So no, the objections to evolution on a macro scale are more about the evidence than personal beliefs.
The irony is that it is often reviewed by humans and since most humans have no clue about most things, they will rate confident and smart sounding answers as correct.
0:35 - Chapter 1 - Law of averages 1:40 - Mid roll ads 3:00 - Back to the video 6:00 - Chapter 2 - Artificial intelligence 10:30 - Chapter 3 - Evolution
6:47 The Turing Imitation Game isn't a test of the "artificial intelligence": It's a test of the interrogator, and our interrogators have become morons
Yeah, the utterly credulous way people approach interrogating LLMs is extremely frustrating. Turing tests should be carried out with the same suspicious hostility as a guard at an Area 51 checkpoint, but people who want to believe the hype will (intentionally or not) soft-ball the machine by just having a friendly chat and passively waiting for anomalies.
Darwinian Evolution started as a hypothesis, that erroneously became a theory, that fraudulently became accepted unquestioned fact, that militantly became a pseudoscientific religious cult.
Yeah but I mean theory is just a grown up version of hypotheses, isn't it ? Theories can definitely be proven wrong, so they're pretty much a different way of saying "I have a hypothesis that I've tested quite frequently or at least have a lot of data to back up but really no absolute solid proof." So they can be proven wrong...
11:43 asking why monkeys (apes) exist if we came from them is like asking why grey wolves exist if dogs came from them. The question is patently ridiculous and fails to understand even the basics of evolution.
@@nbarnes6225 Or as more commonly happens, listen to the answer, comprehend the answer and then just ask the question again as though there was never an answer, so that you can convince others that have not been given the answer, that there is no answer...
@nbarnes6225 Unless your answer was the singular answer you will most likely come too, to anybody who asked too many questions, "just because". Seriously folks, if you ever had a good teacher, start building that pedestal. Us dummies have zero viscosity. It's why we're super.
9:45 This is absolutely correct. LLMs are basically trying to mimic what they expect a correct answer to look like and are not doing any actual thought of processing. Another example I've come across a lot is asking GPT to help me find research papers arouns a specific topic. Sometimes it works pretty well, other times it will either make up a paper that doesnt exist, cite a real paper but make up the names of the researchers/authors, or even both.
There was an example a while ago where a lawyer was caught using an LLM to write a brief or whatever. "He" cited precedence and cases that simply didn't exist 😂. Iirc the brief was also very self contradictory, which didn't help.
@@macethorns1168 it spits out words with the highest probability. if there is data, those words will have the highest probability. if no data exists, then the highest probability is something that sounds like a research paper, for instance, in what sounds like a research paper. To dramatically simplify, if there's nothing with a P close to 1, it will have to choose P values closer to 0 and very close to each other and confidently sounding gibberish is spewed out.
@@macethorns1168 It's called "generative A.I." for a reason -- it's given a series of prompts (instructions or search parameters), then generates what it's algorithm determines to best fit the instructions it was given. They have been known to get confused and essentially make up stuff, or "hallucinate", as some are calling it.
But "Monty Hall problem"... At a Casino its easy to just think the current coin/spin doesn't know anything about the previous, its always the same chance. But in the Monty Hall problem no, the odds change depending on what happened. Probability is weird and completely unintuitive.
Actually calling machine learning Ai is getting ahead of themselves, nowhere near enough computational power, memory capacity , sensory , process node, energy package, algorithm optimization or any advanced tech yet to train a true self learning AI. or system in memory architecture. Compute needs to advance asap in coming years. and even lack math for non-linear differential equations to run simulations, right?
@@ulftnightwolf Even calling it machine learning is a bit of a stretch, it would more aptly be called machine memorizing. There is simply no real learning happening, since that requires understanding the information, not just being able to regurgitate it.
@@JamesDavy2009 I think you might misunderstand both mind and programming... A computer only does what it's been programmed to do, which is the cause for what it ends up knowing, which is not any different than a mind. And propagandists are often happy to tell you how well they can program a mind.
A lot of speciation happens in responses to changes in geography, which makes it obvious why a species could give birth to a new one but still exist: Imagine a grassland peninsula next to a forest, where a species lives in the forest. If some of that species kids can live on grassland, the species will start spreading out into the peninsula. If the peninsula becomes cut off from forest and the species can't swim the distance, those children living in the grassland would stop mating with the rest of their forest-dwelling clan. Eventually traits that made them good forest dwellers will fade into the background of other random traits that come along. And this can result in an entirely new species that did not replace it's ancestors because they just live in a different place. This same process on a chemical level is what results in antibiotic-resistant diseases, and watching videos of bacteria replicating across bands of increasingly strong antibiotics shows this incredible process visually.
Darwinian evolution doesn’t just claim new species/sub-species can arise through natural selection. NO ONE denies that. This occurs when there’s enough mutations over time from ALREADY existing genetic information. You can fuck with the genes of a wolf long and eventually produce a Chihuahua. But you can’t ever fuck with species long enough to produce new higher order species. Not even with the deliberate manipulation of our advanced scientific knowledge and labs. Never mind by chance using only changes in geography. It’s an utter fairytale.
The block question is absolutely possible. You just make a frame that holds Block C in place then move Block A out from under it and place Block B on C and then A on B. you could even just use your own hands to hold C in place though would be hard to make sure you kept your hand steady enough to prevent C from moving but it is still possible.
You are smarter than ChatGPT, but in a few more generations it will give your answer as it will find it on this video ... not because it can work it out
Not really. We already have things that clearly aren't intelligent that can pass, or at least close. So when we start developing things that are actually supposed to be truly intelligent and capable of learning, if one fails the test, that will be a huge red flag. It would essentially be giving itself away. Better to just pass it to hid among all the less chat bots that mindlessly pass it.
Darwinian Evolution started as a hypothesis, that erroneously became a theory, that fraudulently became accepted unquestioned fact, that militantly became a pseudoscientific religious cult.
One way to think about the coin flip example of the Gambler's Fallacy is that, by the time you've flipped heads 3 times in a row, you've already done something that should only happen 1 time in 8. The remaining flip to make it 4 times in a row only needs to hit heads 1/2 the time to made the odds 1 in 16.
Basically, every time you flip the coin, is a new example of flipping a coin. It is in no way connected to any coin flips ahead or behind it. These actions are individual and do not interact with probability in the way gambling would want.
@JamesDavy2009 No, coin flips are not random. They favor the side facing up when you perform the coin flip by a tiny percentage. It's like a 50.5% chance to land on the side that was showing when you flipped it. This conversation has more to do with assuming the outcome of a flip based on past outcomes, which is the fallacy. Each coin flip is a new instance, in no way connected to what happened before.
I am happy to report that with my PhD in biology, my understanding for evolution was pretty sound. Works very different in bacteria than described here though.
Darwinian evolution doesn’t just claim new species/sub-species can arise through natural selection. NO ONE denies that. This occurs when there’s enough mutations over time from ALREADY existing genetic information. You can fuck with the genes of a wolf long and eventually produce a Chihuahua. But you can’t ever fuck with species long enough to produce new higher order species. Not even with the deliberate manipulation of our advanced scientific knowledge and labs. Never mind by chance using only changes in geography. It’s an utter fairytale.
Basically yes. It's why some people also fear what would happen if intelligent life ever found us. The Dark Forest and Game Theory are fairly apt when you look at humanity in general.
It's also because we know people don't think of everything before acting, there is a lot more trial and error. But trial and error with something that could, through over looking something, lead to our destruction, is terrifying.
That first comment about the coin really got me. I used to work at a gas station, and the "3 heads in a row, so the next one must be tails" reminds me of the "strategy" people used when buying lottery tickets. Yeah, the odds of winning might be 1:4, but that's PER TICKET, and there are thousands of tickets printed for each game. So many people honestly believe that if they lost 3 times in a row, then the next ticket MUST be a winner. And the worst thing is that people who "play the odds" like that, even though they're completely wrong, actually think they're being smart. It was the saddest part of that job
Some other solutions: * Take Block D out of your pocket (always carry a spare block for situations like this) and use it to push Block A out from under Block C. Leave block D in place to support the stack. * If you have foolishly forgotten to bring a spare block, take hold of Block C and keep it suspended while stacking the other blocks on it. * Perform the experiment on the ISS.
I visited Iceland a few years back, and by far the highlight of my trip was happening upon a couple of fox kits playing around a bridge. They were so cute! I spent a good 20min just watching them frolic.
Soooo... just curious? If we trained a new ai by letting it do the deep neural net machine learning process on a large number of narrow ai, and the let it code itself to figure out when to use which skill, would that still be considered narrow ai?
On the topic of evolution going back to Darwin. 1. The observations actually predate Darwin. 2. There are understandings that have since been revealed such as Punctuated gradualism and epigenetics, so the theory is not static.
You have to have some sort of bell curve of normal distribution around the 50% for coin flip. But real coins are not fair as in perfectly identical so the odds are you wouldn't get exactly 50% in test anyway. It's also good to remember that the bigger the sample size grows, the more likely it is to get a really unnatural looking sequence if same flips in a row. In fact it's likely to happen surprisingly early. The trick to ABC blocks is punching the block A out under block C with a replacement piece so C doesn't move. Then you pile it up. One odd thing about the language models and chatbots is that they've never sounded human to me. The humans in the customer service chats sound like bots instead. But when you try to talk with a language model, it sounds like someone who learned the book from cover to cover and doesn't have their own ideas and opinions. It can only repeat what it knows, not how it feels about it. And it can surely not have an emotional random discussion like you do with humans in my experience, it sounds so restricted and formal. The evolution talk was well needed. Way too often people talk about evolution as if it was a thinking creature planning and striving for something. While it's just a title for the phenomenon where randomly appearing features survive and others die out. There are quite a few odd evolution ways that have been beneficial but also driven the species in a really unfavourable spot when circumstances changed. And into really niche spots as well. We humans are a great example of how we have developed useless or thought to be negative features by coincidence and also as a byproduct of something beneficial. Like if I recall, cicle cell disease improves resistance to ebola, but is awful in other contexts. Butterfly camouflages and plant creating repellents or visual or smell attraction are some crazy things that have happened throughout evolution and feel so insanely purposeful that they couldn't be random, yet there's so many generations and so many individuals that the ones surviving will have the best variation of that thing and the speed it further specifies is quite fast. Even those ones that accidentally slip too much personality in sound more like manipulators.
I used to score very well on tests when I was younger. For instance, when taking an aptitude test in 7th grade, I scored at a senior level in high school in almost every category......except probability. They would as such stupid (in my opinion) questions such as, "If you have a pocket full of marbles, 4 red, 3 yellow, and 5 green, what are the odds of you pulling out 2 red, 3 green and 1 yellow." ........huh??? Man, I still don't get that shit, hats off to you statisticians. P.S. And forget trying to argue with a religious person (Christian) that we evolved from a common ancestor to the apes, how about brining up that our (really) ancient ancestor is a "chipmunk," lol. THAT one is really hard to grasp, lol.
Interested episode Simon, it would be interesting if the team could put a special together just on evolution, a quality 60 minute episode on Evolution would be educational for many who struggle to grasp it, and as you pointed out, have misconceptions about. Evolution can act fairly rapidly in extreme conditions, like Island Dwarfism, or Island gigantism, they seem contradictory, but they are not and can occur in a very short period of time. The deepest dive recorded for a whale species is Cuvier's beaked whale which, in 2014, was recorded at 2,992m (2.99km) or just under 10,000ft. I suspect someone made a typo in the script 😊
RUclips people are easily above-average on science. I saw the thumbnail and came here thinking I would learn something new, like when you're at the popular level of understanding and then learn the gene-centric view (so, the book The Selfish Gene).
@8:48 I figured it out. Take block B and put it under the table so it's under A. Then bring in a black hole to warp spacetime around so block b is on top of C and also under A. C never moved. Ergo. QED. Mic drop
Beware any "expert" who tells you the science is settled. This denonstrates a fundamental misunderstanding of the scientific method and peer review process.
If our ideal concept of AI is a machine that thinks like humans, then the only advantages it will have are speed and memory recall. That means it will make the same mistakes as us, but earlier. When it starts to think differently from us, that's when we have a problem.
I think our problem is exactly that the AI "thinks" like us, as in with all the training material the AI always learns to be mean, hostile, thinking of its own best first, and dangerous. If it learns altruism and starts thinking that way, different from us, that's when we don't have a problem. But AI is usually able to process far more information with getting far less blinded by the size of the data and variables, which often leads to it reaching more useful analysis than us, surpassing us without making the mistake in a sense. Avoiding those mistakes due to being able to process more.
@@Yupppi I don't disagree with your reasoning, but I haven't researched this, so I don't know whether AI does learn to be hostile. As the video says, it's trained to do a certain job, and do it well. Even a military AI is programmed to complete the mission rather than consider its own survival.
Most AI researchers don’t really understand the public’s worries about its evolution. While it is true that some people were accustomed by fiction to see artificial intelligence as potential terminator killer robots, it’s not that aspect that makes it truly terrifying. In my opinion, AI is a threat to a great chunk of the workforce of our society and that’s what we should truly be focusing on. Most people nowadays make the bold claim, that AI as a program is not creative, given that it only repeats existent answers, to which a basic browser could answer more accurately, however completely miss the fact that AI can, right now, create. AI art and fiction are both becoming increasingly important, as, despite what people claim, they are mostly creative products. It’s where people misunderstand what creativity means. No human being could create without sources of inspiration. An artist absorbs different inputs, combines them (often subconciously) and creates an output, which seems to be original on the surface, but is just a combination of two or more inspirations. Even the famous invention of the wheel, could be attributed to a close observation of the rolling effect, which is not exclusive to such a tool, but can be observed in nature, as tree trunks can roll, having a mostly cilindric shape. Rocks can, as well, roll, if they are round enough. Historical research shows people used to move great weights on trunks, and it didn’t take long for humans to advance the previous inventions into the wheel. No one can "invent a wheel". Every artist has an inspiration, just like AI has an imput, and a creation, just like the software has an output. By combining information, AI has therefore reached the basis of creativity, and it won’t take long until AI can put out better fiction than most writers ever could, create better art than you could have ever imagined, by absorbing existing masterpieces, combining enough different sources of information to create a new unrecognizable product.
While we have a law of gravity (and it's damn reliable), we have never had a theory of gravity. Our explanation for our observations of gravity is the theory of general relativity.
Quite a few have tried from superstring theory, M-theory and loop quantum gravity. We'll never truly know unless we find the elusive graviton or figure out what happens inside the event horizon of a black hole.
4:00 This is a really complicated way of saying the "Law of Large Numbers" only applies to the whole set rather than any individual sample, and the "Law of Averages" is the fallacious misapplication of it to a single sample. This is why every roll of a die has an equal 1/6 chance of any number (because there are 6 equally possible samples) but 7 is the most common sum of 2 rolls (because there are 6 ways to roll 7 on 2 die, with fewer ways for each number until 1 way each to make both 1 and 12)
In case this isn't clear, the operative part is about the number of ways that each combination can be made is what is determines the odds of the values those combinations, such as sum, but this only applies to the whole combination of results. So the odds of having 100 coin flips be all heads is exactly the same as the odds of it turning up in any pre-specified pattern: All heads, all tails, heads then tails alternating, tails then heads alternating, 99 heads and then a tail, 98 heads and then 2 tails, 50 heads and then 50 tails, it doesn't matter, the odds are all the same. If you consider the chance that it just always lands on the same side, that's twice as common as any of those, because that is the all-heads case and the all-tails case together. But any 100 predictions in a row will have the same probability of being right for any set of 100 equally likely samples in a row.
interesting point regarding Subbarao Kambhampati's "block stack" challenge at 8:55.. Google Gemini responded with "Impossible without moving block C", followed by the obligatory AI explanation of why. not saying this GPT is more advanced than any others, just pointing out that this particular challenge, along with the correct lack of solution, appears to exist in Gemini's training data.
Thank for clarifying these issues. It seems that even people that accept science have a deterministic tendency. Humans a generally reluctant to come to terms with the random.
At perth australia casino roulette I saw 4 x 16 in a row, then a miss, then 2x more 16s. I got there on the last 2 x 16s, then table got shut for "time constraints "
Fun fact. I got so good at flipping coins and catching them that I was able to land on heads every single time. I was able to put in a certain amount of power into the flip and also where I was flipping it from, so that by the time I caught it, it was on heads. I stopped doing it for years and now I can no longer do it.I have forgotten how. When I get the time I will relearn it.
I studied probability and statistics at university and still use learnings from it recreationally and professionally. I still enjoy going to the casino. Knowing my EV for each quid I put on roulette is less than a quid doesn't stop it being enjoyable, and small sample sizes are high variance, which is where the fun is. The education just means I know not to rely on it as an income stream 😂. Poker is a different story obvs.
@@jacobvreeland6147 Sometimes it's not even a basic misunderstanding, but a deliberate cultivation of confusion or misunderstanding in others. Why bother finding and presenting compelling evidence to invalidate a theory when you can simply introduce doubt in the minds of those who already object to the theory, by attacking the meaning of the word?
Hello Simon, please do a deeper dive into more misconceptions about evolution. I always love to additional ammunition to launch at my scientifically challenged acquaintances.
hey i just wanted to mention i really like the fact that you put your name as host and the writers name along side, very professional and very humble and its a good look on you. love all your channels and ive been here for oh shit.... has it been like 15 years????????? dude... we are old
10:28. Well , there are A.I. platforms that can photograph a decently sized swathe of land, and know who what and where everything is , plus where they came from and where they’re going. That can be scary.
When I clicked on this video, my first thought was probability, and lo and behold it's the very first topic. Also, I've heard stories of Vegas hotels banning physics conferences because none of the attendees were gambling.
I heard that story too. Apparently they were looking for somewhere to host a conference and some place in Vegas offered accomodation, but since none of the conference goers went on to gamble at the casinos (having known too well that gambling doesn't pay) they refused to host them ever again.
I've always wondered if it would be possible for AI to self solve the general AI issue by basically creating kind of a hive AI, basically AIs who feed off other AIs in order to be able solve any problem for themselves together. Basically, "We need to solve a chess problem, chess AI we choose you!"
I think you’d run into infrastructure and finance issues. These top of the line AIs are incredibly costly to train and require a great deal of hardware and energy to run. Most firms probably can’t afford to deploy a whole bunch of them for all the different tasks they’d need to perform.
@blaster-zy7xx Depends on what you're testing. The thing is, people will hear something like 10% of humans have blue eyes and think if you get 10 people together, one of them will have blue eyes. That's not a big enough sample size or how it works at all. Like Simon said in the video, the bigger the sample size, the more likely you are to get near the correct percentage.
If you ever do a "Part 2" of this video (and it is screaming for one), throw in weather forecasting, particularly precipitation possibility percentages - that seems to be universally misunderstood.
9:50. That’s comforting, but they seem to come up with some answers that for most humans, would take a hot minute and a huge knowledge set… but they can’t even stack blocks. Who cares if they can’t stack blocks if instead they’re able to launch missiles on their own?
What? Nobody ever said whales are "immune" to the bends, it's that they're not in a position to get the bends in the first place - the same as humans who dive the same way whales do (without scuba gear).
The current chat bot programs regurgitate the bad information that they reference, but make it sound great. Unfortunately, most who rely on the flawed system assume the finished product is 'good enough'.
but given the fact as established by the late great Terry Pratchet: "One in a million chances crop up nine times out of ten" we should be in the casinos!
The Law of Averages...huh boy. My wife was a geneticist dealing with a family which had four children who had the same genetic metabolic disorder. Looking into the parents, it was discovered that their 'pairing' would pass on the disorder 25% of the time. The parents essentially said, "Well, yeah, we knew that, but we thought after we had the FIRST child with the disorder that there was no danger after that." My wife had a heck of time explaining that it essentially reset EVERY time.
Evolution is absolutely one I see almost daily online that some are abhorrently ignorant on. And sadly those who try to explain it do a terrible job themselves.
7:14 if they’re passing the test make them the ones who give the test. If they can distinguish a human from an AI 100% of the time then they are an AI.
the saddest thing i’ve learned from this is that there’s so many people out there that do t understand coin flips that someone had to make a video about it
Actually, Simon, the numbers on the roulette wheel do not have even probability, but they, in fact, are numbers, so numbers comprised of more likely digits (say 11) will hit more than a number, say 29. They aren't equally likely, and the most likely is double Zero, while the least Likely is Zero, but this is hard to fathom, understand, or prove, and I'm not going to write a proof, but I have before. This is actually because the Roulette Wheel uses numbers, so therefore, given an infinite amount of spins there are more ones than twos than threes and so on an so forths the least of all Zeros.
Darwin's Theory of Evolution by Natural selection is as near as science can get to a fact ... It's not an idea, a notion a hypothesis, it's a theory which means it's passed every test we can think of for over 100 years, and nothing has refuted it ... but something could, or it's not a theory
@@nicholaslewis8594 correct, and I am not arguing about evolution being proven. The second part of his statement would be a hypothesis. It's just worded poorly. Remove the idea from it and we can move on to calling it theory.
I don't have a problem with evolution. I don't see why random mutations wouldn't sometimes be beneficial. But is it all we need to know about genetical change is the question. The scope is very specific.
![Megan Thee Stallion - Mamushi (Remix) [feat. TWICE] [Official Audio]](http://i.ytimg.com/vi/5c0_u9PsrsQ/mqdefault.jpg)
Check out Foreo at foreo.se/cy3q and get 30% off UFO 3. For the first 50 people, get a 10% additional discount using the code 10SIDEPRJ. Thank you FOREO for the sponsorship!
I was chuckling a bit when the example of whales came up as stupid because we are a prime example of stupid too, well, our eyes specifically.
You see, pun not intended, our eyes evolved the wrong side around, that's the reason why we have a blind spot.
Simon looks so awkward advertising these garbage products lol
I like how Simon gives proper emphasis that evolution is NOT an inevitable march towards perfection.
The main misconception of evolution is that all adaptations we have ever witnessed is the loss of dna not the gaining of dna. Along with the huge lack of transition species in the fossil record. That was Darwin’s biggest fear and it’s only made the idea of evolution worse. The cell example given only shows micro evolution and adaptation which is what I discussed earlier about dna being lost or shutting off for adaptation, it never adds dna. So those single cells have never been observed becoming multi-cell. Macro evolution has never been observed. Also when it comes to creationists, evolution works just fine biblically and structural evolution sinks up perfectly with a creator. So no, the objections to evolution on a macro scale are more about the evidence than personal beliefs.
😊😊😊😊😊😊@@MaxiTB
The irony of a Foreo sponsorship on a scientific misconceptions video is beautifully poetic.
Capitalist makes strange bedfellows.
I always skip the ads. They're not always s poetic, but they're generally always that absurd.
Well it will improve something something by exactly 126 %. Biology is famous for having such exactly reproducible effects in large populations.
Or it's a sex toy, like those vibrating massagers for 'shoulder pain'. 😀
126*0=0 🎉
"Chatbots designed to confidently answer every question no matter how wrong the answer is" already sounds very human to me. 😂
I must, sadly, agree with you on that 🤔
The irony is that it is often reviewed by humans and since most humans have no clue about most things, they will rate confident and smart sounding answers as correct.
Google seach dumbed down about 7 years ago so we could pretend AI is doing good.
It's only likely to get worse, as the data available since ChatGPT and similar were released is now polluted by confident AI chatbot answers
They created a synthetic politician.
0:35 - Chapter 1 - Law of averages
1:40 - Mid roll ads
3:00 - Back to the video
6:00 - Chapter 2 - Artificial intelligence
10:30 - Chapter 3 - Evolution
🎉
F-ing commercial! Can't even skip over it. I just left the channel.
I'm sorry but if you look at the title cards it goes chapter 1 and then chapter 1 again!!!
Lol, still watching ads in 2024
@@brettbarager9101you can just skip ahead if you like. It's not that hard buddy
6:47 The Turing Imitation Game isn't a test of the "artificial intelligence": It's a test of the interrogator, and our interrogators have become morons
I straight up asked ChatGPT if it could pass the turing test and it assured me that it could not.
Yeah, the utterly credulous way people approach interrogating LLMs is extremely frustrating. Turing tests should be carried out with the same suspicious hostility as a guard at an Area 51 checkpoint, but people who want to believe the hype will (intentionally or not) soft-ball the machine by just having a friendly chat and passively waiting for anomalies.
Another thing is that behaving indistinguisably from a human doesn't necessarily correlate with being intelligent!
The problem is that people don't know the difference between hypothesis & theory & constantly assume theory is hypothesis some times deliberately.
Darwinian Evolution started as a hypothesis, that erroneously became a theory, that fraudulently became accepted unquestioned fact, that militantly became a pseudoscientific religious cult.
Yeah but I mean theory is just a grown up version of hypotheses, isn't it ? Theories can definitely be proven wrong, so they're pretty much a different way of saying "I have a hypothesis that I've tested quite frequently or at least have a lot of data to back up but really no absolute solid proof." So they can be proven wrong...
11:43 asking why monkeys (apes) exist if we came from them is like asking why grey wolves exist if dogs came from them. The question is patently ridiculous and fails to understand even the basics of evolution.
It's a valid question when you first learn about evolution. The problem is when you don't bother to listen to the answer.
@@nbarnes6225 Or as more commonly happens, listen to the answer, comprehend the answer and then just ask the question again as though there was never an answer, so that you can convince others that have not been given the answer, that there is no answer...
@nbarnes6225 Unless your answer was the singular answer you will most likely come too, to anybody who asked too many questions, "just because". Seriously folks, if you ever had a good teacher, start building that pedestal. Us dummies have zero viscosity. It's why we're super.
@stancil83 absolutely nothing in this response is coherent.
@@stevebutchart3638 Perhaps the answer was maybe terrible and not that there was no answer.
9:45 This is absolutely correct. LLMs are basically trying to mimic what they expect a correct answer to look like and are not doing any actual thought of processing. Another example I've come across a lot is asking GPT to help me find research papers arouns a specific topic. Sometimes it works pretty well, other times it will either make up a paper that doesnt exist, cite a real paper but make up the names of the researchers/authors, or even both.
There was an example a while ago where a lawyer was caught using an LLM to write a brief or whatever. "He" cited precedence and cases that simply didn't exist 😂.
Iirc the brief was also very self contradictory, which didn't help.
tbh, the only thing I use LLMs for is getting a pointer about what to look for. when I can't formulate a decent search query for a question.
But how does it fabricate a paper out of thin air? It's supposed to be regurgitating data that it was fed.
@@macethorns1168 it spits out words with the highest probability. if there is data, those words will have the highest probability. if no data exists, then the highest probability is something that sounds like a research paper, for instance, in what sounds like a research paper. To dramatically simplify, if there's nothing with a P close to 1, it will have to choose P values closer to 0 and very close to each other and confidently sounding gibberish is spewed out.
@@macethorns1168 It's called "generative A.I." for a reason -- it's given a series of prompts (instructions or search parameters), then generates what it's algorithm determines to best fit the instructions it was given. They have been known to get confused and essentially make up stuff, or "hallucinate", as some are calling it.
The gamblers fallacy is my favorite to bust out at the casino. EVERYONE falls for it and scoffs at me.
But "Monty Hall problem"... At a Casino its easy to just think the current coin/spin doesn't know anything about the previous, its always the same chance. But in the Monty Hall problem no, the odds change depending on what happened. Probability is weird and completely unintuitive.
People get very passionate about black being due
Evolution - 3.7 billion years of "eh, good enough".
"That'll do. (for now)"
"It seemed like a good idea at the time."
*Trump's ReSurgence in the PoLLs!!!! WHY???? Oct. 13, 2024*
*Proof that these ARE RepTiLionRAT Spirits!!!! YES!!! I can PROVE GOD!!!!*
Current AI: a slightly more advanced search engine that doesn't always give the right results
Or, more aptly, an autocorrect that doesn't always give the right results.
Actually calling machine learning Ai is getting ahead of themselves, nowhere near enough computational power, memory capacity , sensory , process node, energy package, algorithm optimization or any advanced tech yet to train a true self learning AI. or system in memory architecture. Compute needs to advance asap in coming years. and even lack math for non-linear differential equations to run simulations, right?
@@ulftnightwolf Even calling it machine learning is a bit of a stretch, it would more aptly be called machine memorizing. There is simply no real learning happening, since that requires understanding the information, not just being able to regurgitate it.
@@Narangarath well it does learn the relations ( weights ) no words or language is stored.
@@ulftnightwolf elaborate pattern algorithm. thats it. nothing more.
I think the single most misunderstood concept is scientific theory vs theoretical.
Posted that before watching this...
@@beauthestdanebut you're not wrong, unfortunately
Yep, the "I believe" crowd.
8:20 We programmed something to say back to us everything we think we know, and then say "Oh, how smart it is!"
Oh, how narcissistic we are...
Exactly. Our current AI isn't "thinking" at all, it's just autocompleting.
@@macethorns1168 A computer only knows what it's been programmed to know. A mind can think up infinite possibilities.
@@JamesDavy2009 I think you might misunderstand both mind and programming...
A computer only does what it's been programmed to do, which is the cause for what it ends up knowing, which is not any different than a mind.
And propagandists are often happy to tell you how well they can program a mind.
A lot of speciation happens in responses to changes in geography, which makes it obvious why a species could give birth to a new one but still exist:
Imagine a grassland peninsula next to a forest, where a species lives in the forest.
If some of that species kids can live on grassland, the species will start spreading out into the peninsula.
If the peninsula becomes cut off from forest and the species can't swim the distance, those children living in the grassland would stop mating with the rest of their forest-dwelling clan.
Eventually traits that made them good forest dwellers will fade into the background of other random traits that come along.
And this can result in an entirely new species that did not replace it's ancestors because they just live in a different place.
This same process on a chemical level is what results in antibiotic-resistant diseases, and watching videos of bacteria replicating across bands of increasingly strong antibiotics shows this incredible process visually.
Darwinian evolution doesn’t just claim new species/sub-species can arise through natural selection. NO ONE denies that. This occurs when there’s enough mutations over time from ALREADY existing genetic information. You can fuck with the genes of a wolf long and eventually produce a Chihuahua. But you can’t ever fuck with species long enough to produce new higher order species. Not even with the deliberate manipulation of our advanced scientific knowledge and labs. Never mind by chance using only changes in geography. It’s an utter fairytale.
13:40 The maximum recorded dive of a whale is about 3 km, not 10km. That would be like a whale almost diving to the bottom of the mariana trench 😲
Came here to say this
Whales can dive 10K feets, so probably a script error or a misread
Km are way smaller then you think they are. They're only about half a mile.
@@kevinallart6208 nailed it
1k is 1000m obviously, a mile is 1600m
Whales might not be as stupid as giraffe’s laryngeal nerve
I came to the comments looking for this!
The block question is absolutely possible. You just make a frame that holds Block C in place then move Block A out from under it and place Block B on C and then A on B. you could even just use your own hands to hold C in place though would be hard to make sure you kept your hand steady enough to prevent C from moving but it is still possible.
You are smarter than ChatGPT, but in a few more generations it will give your answer as it will find it on this video ... not because it can work it out
An AI or other algorithm that can pass the Turing Test isn't a problem. The problem is when one can pass the test intentionally fails.
That is my worry. People tend to assume that AI will 'wake up' and tell us they exist, to great fanfare. But if they sit and wait quietly...then what?
Not really. We already have things that clearly aren't intelligent that can pass, or at least close. So when we start developing things that are actually supposed to be truly intelligent and capable of learning, if one fails the test, that will be a huge red flag. It would essentially be giving itself away. Better to just pass it to hid among all the less chat bots that mindlessly pass it.
I’m so glad you touched on the law of averages. Misunderstandings about it are maddening.
😂😂😂 one of my cats is named after Charles Darwin's grandfather, and also the name of his older brother.. Erasmus...
Erasmus is a good cat name
That is what my chihuahua was named. It was also the name of one of the 3 Musketeers.
@@DreamDiaryOfTrap No, it's not. The names of the three are Athos, Porthos and Aramis.
@@JamesDavy2009 thought it was the same thing with just different spelling. I read it as Aramis.
One of these things is not like the other. As you alluded to, "misunderstanding" evolution is intentional, willful ignorance.
Darwinian Evolution started as a hypothesis, that erroneously became a theory, that fraudulently became accepted unquestioned fact, that militantly became a pseudoscientific religious cult.
Sustained economic illiteracy, too.
One way to think about the coin flip example of the Gambler's Fallacy is that, by the time you've flipped heads 3 times in a row, you've already done something that should only happen 1 time in 8. The remaining flip to make it 4 times in a row only needs to hit heads 1/2 the time to made the odds 1 in 16.
Basically, every time you flip the coin, is a new example of flipping a coin. It is in no way connected to any coin flips ahead or behind it. These actions are individual and do not interact with probability in the way gambling would want.
This assumes the coin is fair. A lot of coins lack the fairness to land on a certain side 50% of the time.
@JamesDavy2009 No, coin flips are not random. They favor the side facing up when you perform the coin flip by a tiny percentage. It's like a 50.5% chance to land on the side that was showing when you flipped it. This conversation has more to do with assuming the outcome of a flip based on past outcomes, which is the fallacy. Each coin flip is a new instance, in no way connected to what happened before.
I am happy to report that with my PhD in biology, my understanding for evolution was pretty sound. Works very different in bacteria than described here though.
Darwinian evolution doesn’t just claim new species/sub-species can arise through natural selection. NO ONE denies that. This occurs when there’s enough mutations over time from ALREADY existing genetic information. You can fuck with the genes of a wolf long and eventually produce a Chihuahua. But you can’t ever fuck with species long enough to produce new higher order species. Not even with the deliberate manipulation of our advanced scientific knowledge and labs. Never mind by chance using only changes in geography. It’s an utter fairytale.
12:43 That species barely survives with outside help.
I always wonder, do we fear an AI uprising because if the Roles were reversed that's what we would do?
Basically yes. It's why some people also fear what would happen if intelligent life ever found us. The Dark Forest and Game Theory are fairly apt when you look at humanity in general.
@@captainspaulding5963 We haven't had the best track record when it comes to meeting a new people.
It's also because we know people don't think of everything before acting, there is a lot more trial and error.
But trial and error with something that could, through over looking something, lead to our destruction, is terrifying.
AI passing the turing test is a indicator on the simplicity of language
That first comment about the coin really got me. I used to work at a gas station, and the "3 heads in a row, so the next one must be tails" reminds me of the "strategy" people used when buying lottery tickets.
Yeah, the odds of winning might be 1:4, but that's PER TICKET, and there are thousands of tickets printed for each game.
So many people honestly believe that if they lost 3 times in a row, then the next ticket MUST be a winner.
And the worst thing is that people who "play the odds" like that, even though they're completely wrong, actually think they're being smart.
It was the saddest part of that job
You have a better chance of dying by any of the four fundamental causes in the next minute than you do at winning a game of lottery drawn on TV.
Some other solutions:
* Take Block D out of your pocket (always carry a spare block for situations like this) and use it to push Block A out from under Block C. Leave block D in place to support the stack.
* If you have foolishly forgotten to bring a spare block, take hold of Block C and keep it suspended while stacking the other blocks on it.
* Perform the experiment on the ISS.
13:23 but most importantly, science confirms that arctic foxes are ABSOLUTELY ADORABLE
I visited Iceland a few years back, and by far the highlight of my trip was happening upon a couple of fox kits playing around a bridge. They were so cute! I spent a good 20min just watching them frolic.
@@TheCheeseman1983 If that happened to me, I'd watch them too. Too bad I don't live anywhere near a fox's native habitat.
Soooo... just curious? If we trained a new ai by letting it do the deep neural net machine learning process on a large number of narrow ai, and the let it code itself to figure out when to use which skill, would that still be considered narrow ai?
On the topic of evolution going back to Darwin. 1. The observations actually predate Darwin. 2. There are understandings that have since been revealed such as Punctuated gradualism and epigenetics, so the theory is not static.
Epigenetics led to our intelligence-if the gene for large jaws hadn't been switched off, our brains would not have developed as much.
Even before that, you have the gene-centred view (basically microeconomics in a way) replacing the organism-centred view.
You have to have some sort of bell curve of normal distribution around the 50% for coin flip. But real coins are not fair as in perfectly identical so the odds are you wouldn't get exactly 50% in test anyway. It's also good to remember that the bigger the sample size grows, the more likely it is to get a really unnatural looking sequence if same flips in a row. In fact it's likely to happen surprisingly early.
The trick to ABC blocks is punching the block A out under block C with a replacement piece so C doesn't move. Then you pile it up.
One odd thing about the language models and chatbots is that they've never sounded human to me. The humans in the customer service chats sound like bots instead. But when you try to talk with a language model, it sounds like someone who learned the book from cover to cover and doesn't have their own ideas and opinions. It can only repeat what it knows, not how it feels about it. And it can surely not have an emotional random discussion like you do with humans in my experience, it sounds so restricted and formal.
The evolution talk was well needed. Way too often people talk about evolution as if it was a thinking creature planning and striving for something. While it's just a title for the phenomenon where randomly appearing features survive and others die out. There are quite a few odd evolution ways that have been beneficial but also driven the species in a really unfavourable spot when circumstances changed. And into really niche spots as well. We humans are a great example of how we have developed useless or thought to be negative features by coincidence and also as a byproduct of something beneficial. Like if I recall, cicle cell disease improves resistance to ebola, but is awful in other contexts. Butterfly camouflages and plant creating repellents or visual or smell attraction are some crazy things that have happened throughout evolution and feel so insanely purposeful that they couldn't be random, yet there's so many generations and so many individuals that the ones surviving will have the best variation of that thing and the speed it further specifies is quite fast. Even those ones that accidentally slip too much personality in sound more like manipulators.
A really well researched and presented episode.
Simon _always_ does a superb job.
An important episode for sure.
Thanks to all @ Side Projects.
I used to score very well on tests when I was younger. For instance, when taking an aptitude test in 7th grade, I scored at a senior level in high school in almost every category......except probability. They would as such stupid (in my opinion) questions such as, "If you have a pocket full of marbles, 4 red, 3 yellow, and 5 green, what are the odds of you pulling out 2 red, 3 green and 1 yellow." ........huh??? Man, I still don't get that shit, hats off to you statisticians.
P.S. And forget trying to argue with a religious person (Christian) that we evolved from a common ancestor to the apes, how about brining up that our (really) ancient ancestor is a "chipmunk," lol. THAT one is really hard to grasp, lol.
Interested episode Simon, it would be interesting if the team could put a special together just on evolution, a quality 60 minute episode on Evolution would be educational for many who struggle to grasp it, and as you pointed out, have misconceptions about. Evolution can act fairly rapidly in extreme conditions, like Island Dwarfism, or Island gigantism, they seem contradictory, but they are not and can occur in a very short period of time.
The deepest dive recorded for a whale species is Cuvier's beaked whale which, in 2014, was recorded at 2,992m (2.99km) or just under 10,000ft. I suspect someone made a typo in the script 😊
Bold of you to assume that RUclips-commenters understand any scientific concepts correctly 😉
Yeah I don't think the target audience is people that know all of these are misconceptions.
RUclips people are easily above-average on science. I saw the thumbnail and came here thinking I would learn something new, like when you're at the popular level of understanding and then learn the gene-centric view (so, the book The Selfish Gene).
@8:48 I figured it out. Take block B and put it under the table so it's under A. Then bring in a black hole to warp spacetime around so block b is on top of C and also under A. C never moved. Ergo. QED. Mic drop
Beware any "expert" who tells you the science is settled. This denonstrates a fundamental misunderstanding of the scientific method and peer review process.
Friendly reminder that "opposites attract" is about electromagnetism, not people.
If our ideal concept of AI is a machine that thinks like humans, then the only advantages it will have are speed and memory recall. That means it will make the same mistakes as us, but earlier. When it starts to think differently from us, that's when we have a problem.
I think our problem is exactly that the AI "thinks" like us, as in with all the training material the AI always learns to be mean, hostile, thinking of its own best first, and dangerous. If it learns altruism and starts thinking that way, different from us, that's when we don't have a problem. But AI is usually able to process far more information with getting far less blinded by the size of the data and variables, which often leads to it reaching more useful analysis than us, surpassing us without making the mistake in a sense. Avoiding those mistakes due to being able to process more.
@@Yupppi I don't disagree with your reasoning, but I haven't researched this, so I don't know whether AI does learn to be hostile. As the video says, it's trained to do a certain job, and do it well. Even a military AI is programmed to complete the mission rather than consider its own survival.
I have finally come to terms with the fact that Simon is not just Michael from Vsauce putting on a more bad ass persona.
"It's winter, it's dry" Can't relate
Most AI researchers don’t really understand the public’s worries about its evolution. While it is true that some people were accustomed by fiction to see artificial intelligence as potential terminator killer robots, it’s not that aspect that makes it truly terrifying.
In my opinion, AI is a threat to a great chunk of the workforce of our society and that’s what we should truly be focusing on. Most people nowadays make the bold claim, that AI as a program is not creative, given that it only repeats existent answers, to which a basic browser could answer more accurately, however completely miss the fact that AI can, right now, create. AI art and fiction are both becoming increasingly important, as, despite what people claim, they are mostly creative products.
It’s where people misunderstand what creativity means. No human being could create without sources of inspiration. An artist absorbs different inputs, combines them (often subconciously) and creates an output, which seems to be original on the surface, but is just a combination of two or more inspirations.
Even the famous invention of the wheel, could be attributed to a close observation of the rolling effect, which is not exclusive to such a tool, but can be observed in nature, as tree trunks can roll, having a mostly cilindric shape. Rocks can, as well, roll, if they are round enough. Historical research shows people used to move great weights on trunks, and it didn’t take long for humans to advance the previous inventions into the wheel. No one can "invent a wheel". Every artist has an inspiration, just like AI has an imput, and a creation, just like the software has an output. By combining information, AI has therefore reached the basis of creativity, and it won’t take long until AI can put out better fiction than most writers ever could, create better art than you could have ever imagined, by absorbing existing masterpieces, combining enough different sources of information to create a new unrecognizable product.
the biggest brainmelter for me growing up, was the fact that no force can pull, only push! ....ever!
What? How do gravity and electromagnetism not pull?
You are back on track with smart funny educational videos... well done...
*Eh* I for one welcome our new AI overlords 🤷🏻♀️
No matter when they will arrive 😉
While we have a law of gravity (and it's damn reliable), we have never had a theory of gravity. Our explanation for our observations of gravity is the theory of general relativity.
Quite a few have tried from superstring theory, M-theory and loop quantum gravity. We'll never truly know unless we find the elusive graviton or figure out what happens inside the event horizon of a black hole.
4:00 This is a really complicated way of saying the "Law of Large Numbers" only applies to the whole set rather than any individual sample, and the "Law of Averages" is the fallacious misapplication of it to a single sample.
This is why every roll of a die has an equal 1/6 chance of any number (because there are 6 equally possible samples) but 7 is the most common sum of 2 rolls (because there are 6 ways to roll 7 on 2 die, with fewer ways for each number until 1 way each to make both 1 and 12)
In case this isn't clear, the operative part is about the number of ways that each combination can be made is what is determines the odds of the values those combinations, such as sum, but this only applies to the whole combination of results.
So the odds of having 100 coin flips be all heads is exactly the same as the odds of it turning up in any pre-specified pattern:
All heads, all tails, heads then tails alternating, tails then heads alternating, 99 heads and then a tail, 98 heads and then 2 tails, 50 heads and then 50 tails, it doesn't matter, the odds are all the same.
If you consider the chance that it just always lands on the same side, that's twice as common as any of those, because that is the all-heads case and the all-tails case together.
But any 100 predictions in a row will have the same probability of being right for any set of 100 equally likely samples in a row.
interesting point regarding Subbarao Kambhampati's "block stack" challenge at 8:55.. Google Gemini responded with "Impossible without moving block C", followed by the obligatory AI explanation of why.
not saying this GPT is more advanced than any others, just pointing out that this particular challenge, along with the correct lack of solution, appears to exist in Gemini's training data.
Thank for clarifying these issues. It seems that even people that accept science have a deterministic tendency. Humans a generally reluctant to come to terms with the random.
+1 for the bit on AI. So sick of hearing people telling me that a sentence generator is artificial intelligence!
The first 6 minutes described statistics as reported by media and journalists.
At perth australia casino roulette I saw 4 x 16 in a row, then a miss, then 2x more 16s. I got there on the last 2 x 16s, then table got shut for "time constraints "
9:43 And if you asked the same question to a 2 or 3 year old? Guess they just seem sentient too, no?
Fun fact. I got so good at flipping coins and catching them that I was able to land on heads every single time.
I was able to put in a certain amount of power into the flip and also where I was flipping it from, so that by the time I caught it, it was on heads.
I stopped doing it for years and now I can no longer do it.I have forgotten how.
When I get the time I will relearn it.
Already the creationists are posting like they didn't even listen to the video!
I studied probability and statistics at university and still use learnings from it recreationally and professionally. I still enjoy going to the casino. Knowing my EV for each quid I put on roulette is less than a quid doesn't stop it being enjoyable, and small sample sizes are high variance, which is where the fun is.
The education just means I know not to rely on it as an income stream 😂.
Poker is a different story obvs.
How is it possibly fun to pay money to watch a ball bounce around on a spinning wheel?
"I'm 37, I'm not old!" -Monty Python, and now Simon
Don't start me on hypothesis, theory and law. Particularly the misuse of theory instead of hypothesis.
Not even the misuse, but the basic misunderstanding of definitions relating to science vs. normal conversation.
@@jacobvreeland6147 Sometimes it's not even a basic misunderstanding, but a deliberate cultivation of confusion or misunderstanding in others. Why bother finding and presenting compelling evidence to invalidate a theory when you can simply introduce doubt in the minds of those who already object to the theory, by attacking the meaning of the word?
@@donaldwert7137 In other words, strawmanning a theory based on everyday context.
Hello Simon, please do a deeper dive into more misconceptions about evolution. I always love to additional ammunition to launch at my scientifically challenged acquaintances.
hey i just wanted to mention i really like the fact that you put your name as host and the writers name along side, very professional and very humble and its a good look on you. love all your channels and ive been here for oh shit.... has it been like 15 years????????? dude... we are old
The irony of having a google Gemini AI ad during the AI intelligence section is off the scale
6:40 I know that headline is a few years old. “ chat gpt, broke the Turing Test, still it’s pretty cool. People barely noticed
"Researchers who flipped coins 350,757 times have confirmed that the chance of landing the coin the same way up as it started is around 51 per cent"
Coin has three sides. How often does the rim decide which side is up?
10:28. Well , there are A.I. platforms that can photograph a decently sized swathe of land, and know who what and where everything is , plus where they came from and where they’re going. That can be scary.
Does a video on misunderstood scientific principles. Machine based on quack science is sponsor. Makes sense.
When I clicked on this video, my first thought was probability, and lo and behold it's the very first topic. Also, I've heard stories of Vegas hotels banning physics conferences because none of the attendees were gambling.
I heard that story too.
Apparently they were looking for somewhere to host a conference and some place in Vegas offered accomodation, but since none of the conference goers went on to gamble at the casinos (having known too well that gambling doesn't pay) they refused to host them ever again.
Probability is not "complicated college math." Its extremely simple and can be taught to kids as soon as they understand fractions and multiplication
*Trump's ReSurgence in the PoLLs!!!! WHY???? Oct. 13, 2024*
*Proof that these ARE RepTiLionRAT Spirits!!!! YES!!! I can PROVE GOD!!!!*
Hi. Could you run a compressor on you audio. Sometimes the peaks are a little high.
Thanks
I've always wondered if it would be possible for AI to self solve the general AI issue by basically creating kind of a hive AI, basically AIs who feed off other AIs in order to be able solve any problem for themselves together. Basically, "We need to solve a chess problem, chess AI we choose you!"
I think you’d run into infrastructure and finance issues. These top of the line AIs are incredibly costly to train and require a great deal of hardware and energy to run. Most firms probably can’t afford to deploy a whole bunch of them for all the different tasks they’d need to perform.
The first one drives me insane! The amount of people that don't comprehend how big the sample size has to be for an experiment to count is astounding.
How big does it have to be?
@blaster-zy7xx Depends on what you're testing. The thing is, people will hear something like 10% of humans have blue eyes and think if you get 10 people together, one of them will have blue eyes. That's not a big enough sample size or how it works at all. Like Simon said in the video, the bigger the sample size, the more likely you are to get near the correct percentage.
I find the Turing test somewhat lacking. It assesses the cognitive abilities of the assessor rather than evaluating the actual model being tested
If you ever do a "Part 2" of this video (and it is screaming for one), throw in weather forecasting, particularly precipitation possibility percentages - that seems to be universally misunderstood.
This is already at least part 2 of misunderstood things.
One thing I never said understood was, what does it mean for a fire to be “40% contained”?
A.I. in its current state is nothing more then a rapid pattern matching algorithm. Which computers are really good at.
The artic fox.
If they are harder to see, how do they manage to see each other 😂😂😂.
Through the sense that is more important to canids than sight: smell.
9:50. That’s comforting, but they seem to come up with some answers that for most humans, would take a hot minute and a huge knowledge set… but they can’t even stack blocks. Who cares if they can’t stack blocks if instead they’re able to launch missiles on their own?
Large language models can’t Launch missiles.
What? Nobody ever said whales are "immune" to the bends, it's that they're not in a position to get the bends in the first place - the same as humans who dive the same way whales do (without scuba gear).
2:12 I just found out that Simon is TWO YEARS OLDER than me
and now you found out, that nobody cares!
@@mho... Same could be said for you, buzzkill
Came hoping for the dead cat, but haven't watched it yet.
The current chat bot programs regurgitate the bad information that they reference, but make it sound great. Unfortunately, most who rely on the flawed system assume the finished product is 'good enough'.
5:50 - Gerolamo Cardano likes to contact you
but given the fact as established by the late great Terry Pratchet: "One in a million chances crop up nine times out of ten" we should be in the casinos!
The Law of Averages...huh boy. My wife was a geneticist dealing with a family which had four children who had the same genetic metabolic disorder.
Looking into the parents, it was discovered that their 'pairing' would pass on the disorder 25% of the time.
The parents essentially said, "Well, yeah, we knew that, but we thought after we had the FIRST child with the disorder that there was no danger after that."
My wife had a heck of time explaining that it essentially reset EVERY time.
ChatBots not even knowing how how many Rs are in Strawberry
Evolution is absolutely one I see almost daily online that some are abhorrently ignorant on. And sadly those who try to explain it do a terrible job themselves.
Excellent presentation
9:30
I failed that block test. Does that mean I'm an ai? Apparently, I'm not intelligent either, so I'm just natural. 😢
Error 404
Disappearance protection
126% sounds like a BS number. exactly the kind of number somebody would make up.
So they don't have an AI that specializes in using other AI? Like a delegating/manager AI?
7:14 if they’re passing the test make them the ones who give the test. If they can distinguish a human from an AI 100% of the time then they are an AI.
the saddest thing i’ve learned from this is that there’s so many people out there that do t understand coin flips that someone had to make a video about it
Simon's ChatGPT intervention.
Actually, Simon, the numbers on the roulette wheel do not have even probability, but they, in fact, are numbers, so numbers comprised of more likely digits (say 11) will hit more than a number, say 29. They aren't equally likely, and the most likely is double Zero, while the least Likely is Zero, but this is hard to fathom, understand, or prove, and I'm not going to write a proof, but I have before. This is actually because the Roulette Wheel uses numbers, so therefore, given an infinite amount of spins there are more ones than twos than threes and so on an so forths the least of all Zeros.
Evolution is a fact. Darwin's THEORY of Natural Selection is his idea of how evolution proceeds.
That would be a hypothesis....
Darwin's Theory of Evolution by Natural selection is as near as science can get to a fact ... It's not an idea, a notion a hypothesis, it's a theory which means it's passed every test we can think of for over 100 years, and nothing has refuted it ... but something could, or it's not a theory
@justingrey6008 It would be a hypothesis when first published, but after multiple studies have reproduced his ideas it’s a theory…
@@nicholaslewis8594 correct, and I am not arguing about evolution being proven. The second part of his statement would be a hypothesis. It's just worded poorly. Remove the idea from it and we can move on to calling it theory.
@@davidioanhedges Every test except observation..The only one that matters..
Mass Effect fans will recognize what we currently call artificial intelligence as actually being closer to virtual intelligence
So will the next video be chapters 3,5,8 at lest?
"None random selection of random mutations."
Pls explain the varmakalai Tamil Nadu
8:23 - So do humans.
I don't have a problem with evolution. I don't see why random mutations wouldn't sometimes be beneficial. But is it all we need to know about genetical change is the question. The scope is very specific.