that's because the second captcha you do is for verifying someone else's... I always feel a bit creative at these times. I mean, what IS a bridge, really? That tree there, is a bridge from the underground to the sky...
@brianwelch1579 wdym second captcha? I always get that verifying images thing but not text or whatever that is. Tbh I don't remember if it was a bridge one that didn't let me go through unless I selected ones including a wrong one cuz they say either "you haven't selected all" or "here, try another one", but what I certainly remember is that either way I failed the bridge one which I believe the one that showed a roof of a gas station or tunnel or something was considered one of them. The solution I figured out is to just dart a glance at them and not think too much.
Every other day , I see a post about encryption , data theft , dangers of AI , IT scams and what not ......but nobody is telling me what to do to reduce the chances , how to stay ahead of this ....
1. Change passwords regularly (once a month) 2. Use password generators for random password that contains different characters and letters 3. 2 factor authentication (when sign-in: password + text message or email with code) 4. Don’t open random links (yes, hard to know what link is scam and what is not, but there are websites to check if the link is legit)
@@GCKteamKrispy 5. Always be suspicious of what you see online. Ask yourself how information that's being requested of you might be used against you. 6. Think carefully about any media or messages that you want to upload or post. Ask yourself if you would care whether the entire world sees that message/media. 7. Have more than one email address. Use the second address for important things like finances, don't give it away otherwise. 8. Use a password manager, and don't allow the browser to save passwords for you. 9. Your voicemail has, or should have, a password/PIN. Make it longer and more difficult to guess. 10. Your answer to a security question should be completely unrelated to the actual answer to the question, and it needs to be stored somewhere safe, like your password manager or a physical vault. 11. Panicking leads to mistakes and oversights, and hackers know this. If a person/email/popup is telling you something scary, stop and take a breath. Make sure you can think clearly before making a decision.
@@GCKteamKrispy For 3. it's a lot better to use an authenticator app than email or text messages. Text messages are surprisingly easy to hack into if someone really wants to and has enough money and/or connections.
I wanted to watch this video, but that background music isn't so backgroundy. Please, do not use any music on top of your dialogue track. Playing music while you are talking to us is a distracting practice. Thanks.
The arms race between sites and bots is like a gigantic experiment in generative adversarial networks each side forcing the other to improve and get better and better at their respective goals.
I think what the narrator meant by that is that they’re coded to accept whatever a certain % majority of users click. So in those weird situations where you’re wondering if a particular box counts as part of the image, as long as your choice to include/exclude agrees with what the majority has chosen, you get it right.
but it doesn't tell humans and computers apart, and in fairness, it's a gateway that doesn't need to be there. It's an automated login which can only verify your answer if it has worked out the solution itself, thus defeating the purpose. That's why we have logins.
google acquired it to help with machine learning, and made the users do it. this was way before the AI standards today, back when you had to input the data yourself. so ya it will keep evolving
I had one - hcpatcha I think - that loves transport, and would always ask two sets of questions. I had it ask me to identify boats - cue thoughts of "is that one technically a boat or a ship?", before realising that the system wouldn't care about such detail. Then the next question came up. Choose all the ships.
Software engineer here who deals with bots all the time. The biggest thing now is just simply slowing them down and stopping dumb bots. Think of the locks on your house's door and windows. Will this 100% stop a professional thief who *really* wants in your home for something specific? Nope. But it's good enough to stop some random sketchy people walking up and turning the handle in the middle of the day. So you might as well add an alarm as well, with motion sensors. This will discourage and keep away most would-be thieves.
Ah yes the whole battle I had to have with Steam captcha just to get into my account 💀 had to use my phone and cellular data instead of my computer because I just kept failing it and when I got it right it just didn’t accept it
Copy the captcha that requires you to move a puzzle piece left and right. Genshin Impact has it and the amount of users who successfully completed was around 97% as the captcha claims.
Glad to know that the creator of CAPTCHA fails it all the time too. Haha. It seems harder now but we need new architecture for the newer AI agents that can pass this test nowadays
the over all claim that "robots" are getting better at solving these tests is because the users are feeding these robots the very data only makes sense if Google is absolutely awful at protecting it's gathered intelligence ion how humans solve these puzzles (which is unlikely), or that google turns around and resells the data it learned from its users to other parties that actually use it for breaking through by solving these puzzles. the latter seems more likely as it creates a perpetual problem/solution service for itself to sell to both sides of the root issue.
It's because there's software you can use for computer vision and just run it on your pc like blip. Idk how it's trained but they probably have to get the labelled data somewhere But i think that's not really the reason, it's just that google has reached their goals with that particular type of training data, so they're switching it to something else that is useful
the funniest story I ever heard of defeating captcha was copying the entire CAPTCHA and re-displaying it on another site as a gateway behind porn and let other people solve it. lol
The only way to prove identity is by proving only you know a piece of secret information. Almost everything else can be faked or inferred from public data. You could even have a twin show up in person, in the worst case. Doing things without technology in person is an okay approach, but it's not as convenient.
A losing battle, yea. I don't want to be a slave to some algorithm monitor the tilt and acceleration of my phone before I can check the balance on my bank account. It's just like the idiots that designed 'automated' faucets and paper towel dispensers that require you to spend more time and effort waving your hands around than you would have if you just turned on the water.
I think we're a little bit out before we'll see them as equals but I would say the narrative is slowly changing. We might see a Bender unit in the future! 😁
NO, the guy is lying, the computer is not being taught to identify items in pictures. The computer doesn't know if it sees a cat or a car, it just know that the humans reward it when it select one type of picture over another. It's being programmed to get a reward, so it will return answers it think give it a reward, not answers that is correct. The second lie it that it's about clicking the right pictures, no, it's actually how we move the mouse to click, because the movement pattern of a human is just not the same as a computer.
Obviously the assumption is that people will try to answer correctly, so the reward aligns with correctness. For your second point, that’s what Nocaptcha measures, explained at 4:36
When did Apple’s Notes app become an extension of our brains? on.wsj.com/3YZHDjx
ahahaha xevil
It did not. I use Notion.
please. a mac can't even talk to an ipad. you must be thinking of windows
I’m surprised they didn’t mention that Luis is also the founder of Duolingo.
Thought he sounded familiar!
Literally, I was thinking the same thing that he has a very familiar face and voice and that I saw him wearing that same shirt in another video😂
Finally someone raising awareness
Are you a Robot?
"You haven't selected all the bridges"
THAT IS NOT A BRIDGE!!!!!
that's because the second captcha you do is for verifying someone else's... I always feel a bit creative at these times. I mean, what IS a bridge, really? That tree there, is a bridge from the underground to the sky...
@brianwelch1579 wdym second captcha? I always get that verifying images thing but not text or whatever that is.
Tbh I don't remember if it was a bridge one that didn't let me go through unless I selected ones including a wrong one cuz they say either "you haven't selected all" or "here, try another one", but what I certainly remember is that either way I failed the bridge one which I believe the one that showed a roof of a gas station or tunnel or something was considered one of them.
The solution I figured out is to just dart a glance at them and not think too much.
At this point, I have just accepted I am a robot.
:D :D :D
@@bob_412Did… it… just become conscious?
Robot or NPC?
Yes it was nonsense with pictures and letters. I was human but pc made me ai 🙈🤣
beep boop
Every other day , I see a post about encryption , data theft , dangers of AI , IT scams and what not ......but nobody is telling me what to do to reduce the chances , how to stay ahead of this ....
1. Change passwords regularly (once a month)
2. Use password generators for random password that contains different characters and letters
3. 2 factor authentication (when sign-in: password + text message or email with code)
4. Don’t open random links (yes, hard to know what link is scam and what is not, but there are websites to check if the link is legit)
I recommend Bitwarden as a password manager
@@GCKteamKrispy
5. Always be suspicious of what you see online. Ask yourself how information that's being requested of you might be used against you.
6. Think carefully about any media or messages that you want to upload or post. Ask yourself if you would care whether the entire world sees that message/media.
7. Have more than one email address. Use the second address for important things like finances, don't give it away otherwise.
8. Use a password manager, and don't allow the browser to save passwords for you.
9. Your voicemail has, or should have, a password/PIN. Make it longer and more difficult to guess.
10. Your answer to a security question should be completely unrelated to the actual answer to the question, and it needs to be stored somewhere safe, like your password manager or a physical vault.
11. Panicking leads to mistakes and oversights, and hackers know this. If a person/email/popup is telling you something scary, stop and take a breath. Make sure you can think clearly before making a decision.
@@GCKteamKrispy For 3. it's a lot better to use an authenticator app than email or text messages. Text messages are surprisingly easy to hack into if someone really wants to and has enough money and/or connections.
I wanted to watch this video, but that background music isn't so backgroundy. Please, do not use any music on top of your dialogue track. Playing music while you are talking to us is a distracting practice. Thanks.
The arms race between sites and bots is like a gigantic experiment in generative adversarial networks each side forcing the other to improve and get better and better at their respective goals.
That's a good way to put it
if there's no wrong answer for the image captchas, why do i fail them?
I think what the narrator meant by that is that they’re coded to accept whatever a certain % majority of users click. So in those weird situations where you’re wondering if a particular box counts as part of the image, as long as your choice to include/exclude agrees with what the majority has chosen, you get it right.
but it doesn't tell humans and computers apart, and in fairness, it's a gateway that doesn't need to be there. It's an automated login which can only verify your answer if it has worked out the solution itself, thus defeating the purpose. That's why we have logins.
Fun fact: humanity wastes about a total of 500 years per day on CAPTCHAs.
Reminds me of John Mulaney joke.
"We spend most of our day telling robots that we're not robots, just to look at our own stuff."
The main reason why I have switched over from google to bing. Captacha kept on showing up every time.
Get firefox
Privacy respecting search engines: Qwant, DuckDuckGo, Brave, SwissCows.
Privacy respecting browsers: hardened Firefox, Mullvad Browser, Brave Browser.
@@__________________________bub0 Firefox is not a search engine...
@@__________________________bub0 Firefox isn't a search engine.
fr tho, i kept checking boxes in captcha and i couldn't get past it 💀
had to quit signing in and live without that acc
Oh no 😂 they aren’t that hard though 😂😂
@ppploan7625 it was one of those where i had to select parts of an image
Nice try robot
Are you sure you aren't a robot?
google acquired it to help with machine learning, and made the users do it. this was way before the AI standards today, back when you had to input the data yourself. so ya it will keep evolving
U can use extensions like Buster which are Captcha solvers
It has never been so easy so solve Captchas
In other word, a robot to solve Catpchas.
I am confused. If users’ inputs are used to train big tech’s bots, whose bots are ‘attacking’ big tech’s that make things more challenging?
Google's evil twin, Ogel
Big tech often release their papers or software as open-source. People use this knowledge and software for their own purpose.
Luis also founded duolingo
I had one - hcpatcha I think - that loves transport, and would always ask two sets of questions. I had it ask me to identify boats - cue thoughts of "is that one technically a boat or a ship?", before realising that the system wouldn't care about such detail.
Then the next question came up. Choose all the ships.
I love it when they tell me I'm not human.
In the future: Why proving you are not a human is harder then ever
I leave sites and don't do captcha type junk unless it's obs fake or just passes me through.
The captions (not captchas) don't match the audio after about halfway through. Please fix, it's very difficult to follow and distracting!
Scary thought that the “correct” answer is the one that the majority of respondents has decided upon….
Ref the traffic light test
There are more innovations in detecting robots than an innovation in creating robots.
Google acquired captcha: now I understand why captchas have become ridiculous.
Software engineer here who deals with bots all the time. The biggest thing now is just simply slowing them down and stopping dumb bots.
Think of the locks on your house's door and windows. Will this 100% stop a professional thief who *really* wants in your home for something specific? Nope.
But it's good enough to stop some random sketchy people walking up and turning the handle in the middle of the day. So you might as well add an alarm as well, with motion sensors. This will discourage and keep away most would-be thieves.
Ah yes the whole battle I had to have with Steam captcha just to get into my account 💀 had to use my phone and cellular data instead of my computer because I just kept failing it and when I got it right it just didn’t accept it
Copy the captcha that requires you to move a puzzle piece left and right. Genshin Impact has it and the amount of users who successfully completed was around 97% as the captcha claims.
The video doesn't tackle the clickbait question on the thumbnail.
It says it does count as is what most people click, so not selecting also works, there are more than 1 solution.
The first time captcha was introduced, I strongly believe that they are also used to train bots or AIs that time.
Captchas are indeed getting tricky, making us feel like robots! 😅
Okay bot
Bro you are a bot
If it wants to know if I am a robot then I generally give it a miss unless it is essential to proceed.
Glad to know that the creator of CAPTCHA fails it all the time too. Haha. It seems harder now but we need new architecture for the newer AI agents that can pass this test nowadays
no, it does not. if you get an error if you try to do the sound capcha then google has pre determined that you will fail the first click test.
the over all claim that "robots" are getting better at solving these tests is because the users are feeding these robots the very data only makes sense if Google is absolutely awful at protecting it's gathered intelligence ion how humans solve these puzzles (which is unlikely), or that google turns around and resells the data it learned from its users to other parties that actually use it for breaking through by solving these puzzles. the latter seems more likely as it creates a perpetual problem/solution service for itself to sell to both sides of the root issue.
It's because there's software you can use for computer vision and just run it on your pc like blip. Idk how it's trained but they probably have to get the labelled data somewhere
But i think that's not really the reason, it's just that google has reached their goals with that particular type of training data, so they're switching it to something else that is useful
It has an existential aspect. You try to find out what is a human at the most fundamental level
I haven't paid taxes in a year simply because the IRS website has too many factors of authentication.
2029 Captcha: Solve this quantum calculus differential algebraic topology question to prove that you are not a robot.
As a human, I find it’s hard to prove to a machine that I am a human
Like they can't get a robot to tilt the phone or walk a few steps in one direction??
I feel as though a robot could better prove it’s a person and then I could
..and that's what makes us more human! :)
Yay Skynet!!
Oh, it’s sooooo hard!!!
User displayed captcha... "What the F..."
User passed.
let me guess. bots are getting smarter than us
I sure hope that in the age of self-driving cars ai is able to visually detect traffic lights and crosswalks
Interesting
I wonder what's the captcha method for blind people.
I've given up on certain sites as I can not prove that I am not a robot...
5:08 You’re just going to glance over the privacy invasion of ‘looking at your internet history’?
the funniest story I ever heard of defeating captcha was copying the entire CAPTCHA and re-displaying it on another site as a gateway behind porn and let other people solve it. lol
So this is the guy responsible for my *mild* annoyance
..that’s a traffic light, that’s a bike, kmn, im still not a robot.
I'm not smart enough to pass a lot of these CAPTCHAS. The robots are winning. We're doomed.
Free AI training from us measly humans.
There are many humans Ive run into that couldnt prove they're human.
Yep. I was once banned on Google for typing over 200WPM
whatever, bot 🙄
@mike74h lol. I'm not saying I type everything at that speed. Just searches I've done in the past
First off: we all DON'T have mobile/smart/cell phones.
I always thought captchas were just stupid and i still do.
Eleven times reCaptcha makes me solve before it will let me pass. Motorcycles are the worse, they always fail.
The only way to prove identity is by proving only you know a piece of secret information.
Almost everything else can be faked or inferred from public data. You could even have a twin show up in person, in the worst case. Doing things without technology in person is an okay approach, but it's not as convenient.
We might just become robots at this point
Why are we training robots to do the one thing we specifically don’t want them to be able to do? Does nobody see the flaw in this logic?
How many N's are in mayonnaise?
I haven't seen one AI. I can do that
"There are two N's in mayonnaise."
At least that's what Gemini said.
I quit making a new account in Minecraft because the test was so hard.
I am not 😢
Is the speaker an AI voice?
A losing battle, yea. I don't want to be a slave to some algorithm monitor the tilt and acceleration of my phone before I can check the balance on my bank account. It's just like the idiots that designed 'automated' faucets and paper towel dispensers that require you to spend more time and effort waving your hands around than you would have if you just turned on the water.
Well I guess I’m a robot now
Am I the only one that intentionally misses the difficult parts? No free training from me, and it still lets me pass 😈
Omfg, I hate these. Isn't the handlebar or fender part of the freaking motorcycle!!?!!?
I am not a ROBOT.
I am a BOT. 🎉.
😮
Am i a bot?
No, you aren't selling crypto unwarranted 😂😂
why? captcha was made by something who isn't human
fingerprints like with Apple??
Why is Alan Turing everywhere
it's explained in vdo around 6 minutes
So they wasted our time and sanity to train their models for free.
Them Catcha AI aren't so I. Probably human will retry and retry but bots do not
there must be another way lol
🦉
what if AI watches this video... or the AI can understand what video wants to say?
The AI might have a hard time because the captions don't match the audio after about halfway through.
Skibidi robots
Image a future captcha like:
"In order to register your 🍆hub account, shake your phone in your right hand"
🤦🏾♂️
First too.
So, humans and computers aren't that different to computers now. When are we going to start seeing them as equals.
I think we're a little bit out before we'll see them as equals but I would say the narrative is slowly changing. We might see a Bender unit in the future! 😁
Why can't Aïzheimer recognize a fire hydrant and why aren't we paid to assist his early senility? ;)
Dude talk slowly
didn’t know Chris Hemsworth is an AI expert
this is the most useless thing it wasted my hours of time😡😡
What do you mean "has never been harder"? It is f**king hard.
Yes. “Never been harder” means it is most difficult now.
@@rezwhap Yeah, but for some people it's really hard now!
NO, the guy is lying, the computer is not being taught to identify items in pictures. The computer doesn't know if it sees a cat or a car, it just know that the humans reward it when it select one type of picture over another. It's being programmed to get a reward, so it will return answers it think give it a reward, not answers that is correct.
The second lie it that it's about clicking the right pictures, no, it's actually how we move the mouse to click, because the movement pattern of a human is just not the same as a computer.
Obviously the assumption is that people will try to answer correctly, so the reward aligns with correctness. For your second point, that’s what Nocaptcha measures, explained at 4:36
First
ahahaha xevil