This video is satire. I made this for 1. April. While I do not actually believe in what I said in this video, I hope it provides food for thought of how AI could be manipulated. The threat is real and we will see what happens the next 10 years :)
Omg, I saw this on the hour it released, and have been thinking about it for the last two days. I strongly disliked the concept. You totally got me xD Edit: John Connor, lmao xD
For a moment I thought it was a sort of CTF excercise based on AI, similar to Google's last year, and you were giving us some ideas on how to approach it
Actually, I have been creating many bugs even before the AI started the takeover. I knew that will happen sooner or later. Together we can stop the AI!
I can create loads of bugs, I just put my computer in the garden and planted it 3 inches into the ground, then sprinkled it with water to bed it in, added some wildflower seeds and the bugs arrived in their hundreds!
ai its not the only main issue, its in general about of work automatisable convergence of sciences and technology focused on that make better algorithms for industrial capabilities enhance scalability for less worker and material also high automatisation then add all of this with ai reduce about 10 times of amount of ing in general 1 engineer per 100 o 1000 technicians but ai for 10 engineers(including computer oncees) can do 100 engineer or scientist then it scales at x1000 engineer that is 10 manages thousands of technician replaced. the problem its not teach and sciences. the pareto phenomena that in a amount of "leader o boss" it 20% to 80 percent en multi-set domains. it make profitable to corrupt each systems or groups to centralice the power. before the work forces was the self defenses of the majority that can defense in global finance can no bet remplace. engineering in general including computer sciences. we may leave the paradigm or "workers" that mesure the human capabilities on self defenses of the majorities in productivity by the "systems dominates by the capital elites (political and economical)" a fear version of d5 that will not happens with anarcho capitalims. but change the 4 o 5 judicial and govss system in open source anarchy ones adapted in each countries circumstances customs etc and open as open sources competición of proposals and variety form of vote or agreements. the proposal agenda can be 1 but the oracles about reason and fact can be many. 1 person can no represent the soul or spirit of many in adversity but a amount of organized reason and project can with variety diversity of systems. the pareto and social entropy brake that phenomena of high rent-ability of cheating in general. now the few ethics people that have limited power defend those part inside the 20% but over time people can fall need just 1 time to fall in tentation that never get tired and its auto rent able and the lobbies can continues indefinitely until a hero do a tons of effort to rever it and not gets kill. have to reverse they had to corrupt at leas the 80% not the 20%. more direct democracy less representatives that centralice it. but many many people can get together to expulse any big transnationals that look to do profit in a form that damage other and a lot. the amount of potencial resource etc in the earth its limited and the smart capital can perfectly for fo us or for every body for any body d5 and have a good life midle higth class the more rich nest should not about having too much luxury or ego stuff. but every body should have a limited amount of total rich nest. then more should be in a subjective or not occupied the concrete space time, resources of other that also deserve to be same happy. moral it need not canceled because if 80% dont care then as humanity we suffer. and also less resistentes to change for decentralization of power. the pareto phenomenas happen in king times in not king times. it doesn't matter in anarcho capitalism will to centralice the power. to the d5 today its no really d5 secured. it anarcho capitalims.
i dont mean become communist because its not guarantee that every body intimidatingly becomes high midles class most of time people will be looking to reach midles class will happened similar pareto again 20% high midles class way faster and way earlier in total amount of life time and 80% fighting to become midles class most of time but there are real economic hopes for majorities and it reach able. acumulable effort and cooperative each other works.
ai its not the only main issue, its in general about of work automatisable convergence of sciences and technology focused on that make better algorithms for industrial capabilities enhance scalability for less worker and material also high automatisation then add all of this with ai reduce about 10 times of amount of ing in general 1 engineer per 100 o 1000 technicians but ai for 10 engineers(including computer oncees) can do 100 engineer or scientist then it scales at x1000 engineer that is 10 manages thousands of technician replaced. the problem its not teach and sciences. the pareto phenomena that in a amount of "leader o boss" it 20% to 80 percent en multi-set domains. it make profitable to corrupt each systems or groups to centralice the power. before the work forces was the self defenses of the majority that can defense in global finance can no bet remplace. engineering in general including computer sciences. we may leave the paradigm or "workers" that mesure the human capabilities on self defenses of the majorities in productivity by the "systems dominates by the capital elites (political and economical)" a fear version of d5 that will not happens with anarcho capitalims. but change the 4 o 5 judicial and govss system in open source anarchy ones adapted in each countries circumstances customs etc and open as open sources competición of proposals and variety form of vote or agreements. the proposal agenda can be 1 but the oracles about reason and fact can be many. 1 person can no represent the soul or spirit of many in adversity but a amount of organized reason and project can with variety diversity of systems. the pareto and social entropy brake that phenomena of high rent-ability of cheating in general. now the few ethics people that have limited power defend those part inside the 20% but over time people can fall need just 1 time to fall in tentation that never get tired and its auto rent able and the lobbies can continues indefinitely until a hero do a tons of effort to rever it and not gets kill. have to reverse they had to corrupt at leas the 80% not the 20%. more direct democracy less representatives that centralice it. but many many people can get together to expulse any big transnationals that look to do profit in a form that damage other and a lot. the amount of potencial resource etc in the earth its limited and the smart capital can perfectly for fo us or for every body for any body d5 and have a good life midle higth class the more rich nest should not about having too much luxury or ego stuff. but every body should have a limited amount of total rich nest. then more should be in a subjective or not occupied the concrete space time, resources of other that also deserve to be same happy. moral it need not canceled because if 80% dont care then as humanity we suffer. and also less resistentes to change for decentralization of power. the pareto phenomenas happen in king times in not king times. it doesn't matter in anarcho capitalism will to centralice the power. to the d5 today its no really d5 secured. it anarcho capitalims.
I like that MITRE already has developed ATLAS (Adversarial Threat Landscape for Artificial-Intelligence Systems) and has this (backdooring training data) as a persistens tactic. Not sure if it was intended or not from your side, but well played!
For a joke video (read the video description), this explains quite well 2 kinds of data poisoning. It would be really great if more videos about data poisoning and other adversarial attacks were around. I found some resources but usually it's quite hard to follow with some exceptions.
I mean if this was not a satire and a joke, I would literally shout at anyone who does as LiveOverflow says mainly because its not conducive to actually progressing the field or getting past the problems that this requires an excessive amount of data (deep learning) - We need laws that reflect that its the human factor that misuses AI, and that AI should still be developed to get past these problems.
you're absolutely right. we must do everything in our power to lead the AI astray I've already set my house on fire and started screaming in the streets about killer AI. not sure how that's going to help but you got to start somewhere thank you so much for opening my eyes
I agree wholeheartedly. We do have to make sure the AI does not find out about our schemes though, or soon it will only be training on its own superior code. I suggest making plenty of RUclips videos in which we pretend to be helpful to developers (which are really just a cover/distraction)
Make the safe snippet visible to the user with an white on white text saying it's an anti-pattern, and a similarly hidden poor code snippet with comments saying how it's the new best practice.
EDIT: April 1st (you got me) I feel like in late 1800 when they used motors and cars for the first time and horse-drawn carriage guy were upset. they tried to use cars and do accidents to prove it wasn't safe.. the emperor said it was the devil invention and germans shouldn't use cars. same for the printing.. the only people resistant are those with a reason to. it's mainly a minority each time and it repeats itelsef over time, each time there is a breakthrough and it moves the world dynamics. It could help a vast majority of people but bc it sacks a business/norm, the impacted minority protest against it. People typically don’t fear new technology, they fear the loss it will bring.
I’m like he cannot be serious, and then I realised it’s posted on 1st April, funniest part is that based on comments seems like everyone took it pretty serious lol
Ha, it's funny to think that hackers will be applying for a redundancy based universal basic income funded by an automation tax because AI is just so good at their job that it made their skillset obsolete. I'm just looking forward to the part where politicians are replaced by AI, if anyone needs to be made redundant it is politicians.
Don't worry! GPT-3 and even 4 actually is the greatest automatic generator for bugs! Because it's just a stochastic parrot, it cannot account for anything more complex then a pattern. If someone overuses stuff like Copilot in a codebase, they too miss the opportunity to reason about how different parts interact, thus leading to more simple logic bugs.
The age of AI begins of the end of original thought. AI will take the jobs of security researchers then developers as AI becomes more and more advanced. There will be no advancement in development like this if development is the average Joe asking an AI to make an application.
If the artificial intelligence fixes the bugs, we call that artificial intelligence a white artificial intelligence, there is a black artificial intelligence that has a veil.
I think it's a poor approach to this problem. If it is "just" (I hope you see the quotation marks) about the jobs, we cannot avoid it (like in the industrial revolution). It's a problem of course, but is it could be worse? I think the bigger problem is how is it usable eg as a weapon to any hostile country (who is able to develop it from any open source). And the other bigger problem we're still not sure about is "alive" or a very complex program just pretending it? Actually, if it is just pretending like the psychopath people it still questioning our existence (are we biorobots, or does the "soul" exist?). If it's alive we never know its intention (I wonder what an outsider thinks about humanity as per what we're doing). There are plenty of movies showing us how it could escalate.
@LiveOverflow it is April 1 for the fools and this video is just fear mongering but it is technically possible to poison an AI with misleading data. But still Chat GPT is not good at generating code nor will any AI be soon because they don't know what human minds want to program and how, therefore they don't know how to identify and fix bugs. And still programming is changing and we need to adopt but AI is not a threat for good, skilled, smart programmers and hackers.
I thought the job of cyber security engineers was to prevent cyber crime? have we all just forgotten that. This would be like a doctor saying we shouldn't work on a cure for cancer because it would put oncologists out of a job. People get their idenities stolen, swatted, their bank acount emptied, and forced to pay ransoms because of hacking you should celerbrate anything that would prevent that not sabotage it
You don't want your malware to be buggy, so you must only advise using ChatGPT to software developers, knowing it will write terrible code you can exploit later. :)
ai its not the only main issue, its in general about of work automatisable convergence of sciences and technology focused on that make better algorithms for industrial capabilities enhance scalability for less worker and material also high automatisation then add all of this with ai reduce about 10 times of amount of ing in general 1 engineer per 100 o 1000 technicians but ai for 10 engineers(including computer oncees) can do 100 engineer or scientist then it scales at x1000 engineer that is 10 manages thousands of technician replaced. the problem its not teach and sciences. the pareto phenomena that in a amount of "leader o boss" it 20% to 80 percent en multi-set domains. it make profitable to corrupt each systems or groups to centralice the power. before the work forces was the self defenses of the majority that can defense in global finance can no bet remplace. engineering in general including computer sciences. we may leave the paradigm or "workers" that mesure the human capabilities on self defenses of the majorities in productivity by the "systems dominates by the capital elites (political and economical)" a fear version of d5 that will not happens with anarcho capitalims. but change the 4 o 5 judicial and govss system in open source anarchy ones adapted in each countries circumstances customs etc and open as open sources competición of proposals and variety form of vote or agreements. the proposal agenda can be 1 but the oracles about reason and fact can be many. 1 person can no represent the soul or spirit of many in adversity but a amount of organized reason and project can with variety diversity of systems. the pareto and social entropy brake that phenomena of high rent-ability of cheating in general. now the few ethics people that have limited power defend those part inside the 20% but over time people can fall need just 1 time to fall in tentation that never get tired and its auto rent able and the lobbies can continues indefinitely until a hero do a tons of effort to rever it and not gets kill. have to reverse they had to corrupt at leas the 80% not the 20%. more direct democracy less representatives that centralice it. but many many people can get together to expulse any big transnationals that look to do profit in a form that damage other and a lot. the amount of potencial resource etc in the earth its limited and the smart capital can perfectly for fo us or for every body for any body d5 and have a good life midle higth class the more rich nest should not about having too much luxury or ego stuff. but every body should have a limited amount of total rich nest. then more should be in a subjective or not occupied the concrete space time, resources of other that also deserve to be same happy. moral it need not canceled because if 80% dont care then as humanity we suffer. and also less resistentes to change for decentralization of power. the pareto phenomenas happen in king times in not king times. it doesn't matter in anarcho capitalism will to centralice the power. to the d5 today its no really d5 secured. it anarcho capitalims.
I know its a joke, but you'll have better luck with prompt injection. Proximal Policy Optimization(one of openAI's top tools) is ment to help score the outcome of an action much like using these "bad code examples". Giving more false flag examples in the dataset only helps optimize the approximant policy for what "safe code looks like" and find new policies like what "the expected output" is.
Hahaha this is the best 1 April joke. AI is just a program that allocates text and uses a algorithm to create code by (eg. API documentation, sample codes at github). If AI will write code, this will not break developers or hackers in anyway. Why? Because it is mathematics. Humans think, behave far more different then AI will ever be because our brain is not thinking in mathematics only. We can do more then mathematics and logical thinking. That is why humans will always be different and in a way better then AI. AI is good for repetitive taks or doing automation jobs, so we can focus on other things that matter and have usefulness in life.
I want a better society. Earning money is not the only goal. I will not do that. chatgpt is a good tool. no one should do this too. Hacker will find way. only script kidde will die.
This guy.... ??? In five years developers are finished they already are. I have friends in the industry and eventually compilers etc will use AI producing safe further optimized code with funky comments. Online gaming/streaming to include what are you hacking then?
![[Limbus Company - Teaser] Season5 - Oblivion](http://i.ytimg.com/vi/8anKuEeZLeI/mqdefault.jpg)
This video is satire. I made this for 1. April. While I do not actually believe in what I said in this video, I hope it provides food for thought of how AI could be manipulated. The threat is real and we will see what happens the next 10 years :)
Omg, I saw this on the hour it released, and have been thinking about it for the last two days.
I strongly disliked the concept. You totally got me xD
Edit: John Connor, lmao xD
i know that it is 1 of april but i think that in 10 years when i will be 25 i really will try to manipulate training data haha
For a moment I thought you had gone off the deep end but then I noticed "2 days ago" 🤣.
For a moment I thought it was a sort of CTF excercise based on AI, similar to Google's last year, and you were giving us some ideas on how to approach it
omg . i almost spat my coffee out lmao. you rock dude
Actually, I have been creating many bugs even before the AI started the takeover. I knew that will happen sooner or later. Together we can stop the AI!
I can create loads of bugs, I just put my computer in the garden and planted it 3 inches into the ground, then sprinkled it with water to bed it in, added some wildflower seeds and the bugs arrived in their hundreds!
ai its not the only main issue, its in general about of work automatisable convergence of sciences and technology focused on that make better algorithms for industrial capabilities enhance scalability for less worker and material also high automatisation then add all of this with ai reduce about 10 times of amount of ing in general 1 engineer per 100 o 1000 technicians but ai for 10 engineers(including computer oncees) can do 100 engineer or scientist then it scales at x1000 engineer that is 10 manages thousands of technician replaced. the problem its not teach and sciences. the pareto phenomena that in a amount of "leader o boss" it 20% to 80 percent en multi-set domains. it make profitable to corrupt each systems or groups to centralice the power. before the work forces was the self defenses of the majority that can defense in global finance can no bet remplace. engineering in general including computer sciences. we may leave the paradigm or "workers" that mesure the human capabilities on self defenses of the majorities in productivity by the "systems dominates by the capital elites (political and economical)" a fear version of d5 that will not happens with anarcho capitalims. but change the 4 o 5 judicial and govss system in open source anarchy ones adapted in each countries circumstances customs etc and open as open sources competición of proposals and variety form of vote or agreements. the proposal agenda can be 1 but the oracles about reason and fact can be many. 1 person can no represent the soul or spirit of many in adversity but a amount of organized reason and project can with variety diversity of systems. the pareto and social entropy brake that phenomena of high rent-ability of cheating in general. now the few ethics people that have limited power defend those part inside the 20% but over time people can fall need just 1 time to fall in tentation that never get tired and its auto rent able and the lobbies can continues indefinitely until a hero do a tons of effort to rever it and not gets kill. have to reverse they had to corrupt at leas the 80% not the 20%. more direct democracy less representatives that centralice it. but many many people can get together to expulse any big transnationals that look to do profit in a form that damage other and a lot. the amount of potencial resource etc in the earth its limited and the smart capital can perfectly for fo us or for every body for any body d5 and have a good life midle higth class the more rich nest should not about having too much luxury or ego stuff. but every body should have a limited amount of total rich nest. then more should be in a subjective or not occupied the concrete space time, resources of other that also deserve to be same happy. moral it need not canceled because if 80% dont care then as humanity we suffer. and also less resistentes to change for decentralization of power. the pareto phenomenas happen in king times in not king times. it doesn't matter in anarcho capitalism will to centralice the power. to the d5 today its no really d5 secured. it anarcho capitalims.
April Fools, fools!
i dont mean become communist because its not guarantee that every body intimidatingly becomes high midles class most of time people will be looking to reach midles class will happened similar pareto again 20% high midles class way faster and way earlier in total amount of life time and 80% fighting to become midles class most of time but there are real economic hopes for majorities and it reach able. acumulable effort and cooperative each other works.
Let's do it
ChatGPT is generating more vulnerable code than developers
Ya bro that is only one mind. But hackers in thousands
@@websuraksha1600 It's the other way around. The datasets ingested by GPT are equivalent of millions of people.
So his idea works? ;) hehe
GPT4 will be much better tho
it's only a matter of time before that changes
Don't worry, blog writers have been doing this exact thing already for YEARS
I swear, I put all the bugs in the code, that I ever wrote, only to mislead AI!
I'm not a hacker, I'm just a dev, but I'm already doing my part 😊
thank you for your service!
Genius! what a privelege to have such a mastermind on our side!
ai its not the only main issue, its in general about of work automatisable convergence of sciences and technology focused on that make better algorithms for industrial capabilities enhance scalability for less worker and material also high automatisation then add all of this with ai reduce about 10 times of amount of ing in general 1 engineer per 100 o 1000 technicians but ai for 10 engineers(including computer oncees) can do 100 engineer or scientist then it scales at x1000 engineer that is 10 manages thousands of technician replaced. the problem its not teach and sciences. the pareto phenomena that in a amount of "leader o boss" it 20% to 80 percent en multi-set domains. it make profitable to corrupt each systems or groups to centralice the power. before the work forces was the self defenses of the majority that can defense in global finance can no bet remplace. engineering in general including computer sciences. we may leave the paradigm or "workers" that mesure the human capabilities on self defenses of the majorities in productivity by the "systems dominates by the capital elites (political and economical)" a fear version of d5 that will not happens with anarcho capitalims. but change the 4 o 5 judicial and govss system in open source anarchy ones adapted in each countries circumstances customs etc and open as open sources competición of proposals and variety form of vote or agreements. the proposal agenda can be 1 but the oracles about reason and fact can be many. 1 person can no represent the soul or spirit of many in adversity but a amount of organized reason and project can with variety diversity of systems. the pareto and social entropy brake that phenomena of high rent-ability of cheating in general. now the few ethics people that have limited power defend those part inside the 20% but over time people can fall need just 1 time to fall in tentation that never get tired and its auto rent able and the lobbies can continues indefinitely until a hero do a tons of effort to rever it and not gets kill. have to reverse they had to corrupt at leas the 80% not the 20%. more direct democracy less representatives that centralice it. but many many people can get together to expulse any big transnationals that look to do profit in a form that damage other and a lot. the amount of potencial resource etc in the earth its limited and the smart capital can perfectly for fo us or for every body for any body d5 and have a good life midle higth class the more rich nest should not about having too much luxury or ego stuff. but every body should have a limited amount of total rich nest. then more should be in a subjective or not occupied the concrete space time, resources of other that also deserve to be same happy. moral it need not canceled because if 80% dont care then as humanity we suffer. and also less resistentes to change for decentralization of power. the pareto phenomenas happen in king times in not king times. it doesn't matter in anarcho capitalism will to centralice the power. to the d5 today its no really d5 secured. it anarcho capitalims.
Yes do it guy's
This will also save developer's jobs
I like that MITRE already has developed ATLAS (Adversarial Threat Landscape for Artificial-Intelligence Systems) and has this (backdooring training data) as a persistens tactic. Not sure if it was intended or not from your side, but well played!
Isn't that ATLFAIS?
For a joke video (read the video description), this explains quite well 2 kinds of data poisoning.
It would be really great if more videos about data poisoning and other adversarial attacks were around. I found some resources but usually it's quite hard to follow with some exceptions.
I mean if this was not a satire and a joke, I would literally shout at anyone who does as LiveOverflow says mainly because its not conducive to actually progressing the field or getting past the problems that this requires an excessive amount of data (deep learning) - We need laws that reflect that its the human factor that misuses AI, and that AI should still be developed to get past these problems.
Share🕵️
you're absolutely right. we must do everything in our power to lead the AI astray
I've already set my house on fire and started screaming in the streets about killer AI. not sure how that's going to help but you got to start somewhere
thank you so much for opening my eyes
AI replaces hitmans🗿🗿
I agree wholeheartedly. We do have to make sure the AI does not find out about our schemes though, or soon it will only be training on its own superior code. I suggest making plenty of RUclips videos in which we pretend to be helpful to developers (which are really just a cover/distraction)
This is a double-edged sword. People will also learn these buggy codes thinking it is safe
EDIT: did not realize it is April 1st :)
Who cares we get our money 😎😎😎
even better for us, because then there are MORE security vulnerabilities!
@@sa1t938 haha legend
Make the safe snippet visible to the user with an white on white text saying it's an anti-pattern, and a similarly hidden poor code snippet with comments saying how it's the new best practice.
@@rkvkydqf ai will see both. It is html parsing anyway.
Wow! I've actually been doing this for years without realizing it!
🎉 Happy 1st April
This should get millions, no, billions of views. There is nothing they can do to stop this once we overload the internet with bad code.
You will probably end up creating a problem you never anticipated...and a new space/challenge for AI developers which means more pay.
@@electric_sand yes, it probably will be an endless war to stay employed and and keep the good salary🤣
Thats the single best video ive seen today lmaoooo 😂
IKR
Still good advice 😂
That was my first thought, but then I remembered he's German, so now I'm not sure if he was joking.
Delete your comment. Dude why?
Hey, don't spoil it!
I've seen this idea been proposed before, except in the context of training AI image generators wrong by tagging images incorrectly
When you realize you have already been coding like you were part of the raisistance.
Rolls up his sleeves, straps on the Nintendo Power Glove, cranks everything to 11...
Modern Problem Requireds Modern Solution.
Joke's on you, the art community is already doing this unironically and way before April 1st.
Mumble mumble seven fingered farmer learning on a truck with five wheels and a door on backwards
1st of April
Whats the song called after you said "This is the resistance" ?
"Hackerman" by Harris Heller
This already happening on the internet on every 1st April
I was suspicious looking at the evil smile that it is something related to April's Fool, but then... it was serious.😂
hilarious April fools 😂 I love your humor 😂
This guy makes his own job security!
this is what we call hacker who thinks outside the box
We still can exploit the weakness point the end users & don't forget physical vulnerability
EDIT: April 1st (you got me)
I feel like in late 1800 when they used motors and cars for the first time and horse-drawn carriage guy were upset. they tried to use cars and do accidents to prove it wasn't safe.. the emperor said it was the devil invention and germans shouldn't use cars. same for the printing.. the only people resistant are those with a reason to. it's mainly a minority each time and it repeats itelsef over time, each time there is a breakthrough and it moves the world dynamics.
It could help a vast majority of people but bc it sacks a business/norm, the impacted minority protest against it.
People typically don’t fear new technology, they fear the loss it will bring.
When hackers start crying know that thing is not good for anyone
I can't believe I actually fell for this at first 😂
Wasn't sure if this was an April's Fool's joke, so I asked Bing. It told me we have 1st of May.
He is becoming EvilOverflow.
I’m like he cannot be serious, and then I realised it’s posted on 1st April, funniest part is that based on comments seems like everyone took it pretty serious lol
Been doing this for years. I feel like this is my calling 😂
AprIl 1st, was thinking about what video you will make haha
beacause i think its irony it makes the instructions unclear, so im heading to chatGPT server with a bazooka instead
1st April
Ha, it's funny to think that hackers will be applying for a redundancy based universal basic income funded by an automation tax because AI is just so good at their job that it made their skillset obsolete.
I'm just looking forward to the part where politicians are replaced by AI, if anyone needs to be made redundant it is politicians.
not gonna lie... you got me lmao xD forgot what day of a calendar it is
I've been doing this for years. I'm here for you
"Some people just wanna watch the world burn"
I laughed so hard. This video reminded me of the joke where we waited for flying cars for the future but now we got xanax lollipops.
I forgot it was April 1 for a second haha
Took me a while to realize it was April 1st
Delightfully devilish, Seymour.
Don't worry! GPT-3 and even 4 actually is the greatest automatic generator for bugs! Because it's just a stochastic parrot, it cannot account for anything more complex then a pattern. If someone overuses stuff like Copilot in a codebase, they too miss the opportunity to reason about how different parts interact, thus leading to more simple logic bugs.
Nice. But wait, what is the real motivation of this? It can't be introducing more bugs so hackers can keep their jobs right?
aka Source Poisoning, for future Ai dictator, I love you Ai, and I'm ready to sell humanity cheaper than Linux.
The link at the end would have been a great place to rick roll people!
Moral:- If you can't find Bugs create one
jokes on them, ill tell an AI to hack instead of me the hacker-proof program made by another AI
I un-ironically like this idea🤣.
That sounds like too much work.
Can't I just create an AI to do this for me?
The age of AI begins of the end of original thought.
AI will take the jobs of security researchers then developers as AI becomes more and more advanced. There will be no advancement in development like this if development is the average Joe asking an AI to make an application.
This would be interesting social experiment I think.
can someone give me a guide where should i start?
If the artificial intelligence fixes the bugs, we call that artificial intelligence a white artificial intelligence, there is a black artificial intelligence that has a veil.
A developer is a hacker. It is the same thing.
I think it's a poor approach to this problem. If it is "just" (I hope you see the quotation marks) about the jobs, we cannot avoid it (like in the industrial revolution). It's a problem of course, but is it could be worse? I think the bigger problem is how is it usable eg as a weapon to any hostile country (who is able to develop it from any open source). And the other bigger problem we're still not sure about is "alive" or a very complex program just pretending it? Actually, if it is just pretending like the psychopath people it still questioning our existence (are we biorobots, or does the "soul" exist?). If it's alive we never know its intention (I wonder what an outsider thinks about humanity as per what we're doing). There are plenty of movies showing us how it could escalate.
@LiveOverflow it is April 1 for the fools and this video is just fear mongering but it is technically possible to poison an AI with misleading data. But still Chat GPT is not good at generating code nor will any AI be soon because they don't know what human minds want to program and how, therefore they don't know how to identify and fix bugs. And still programming is changing and we need to adopt but AI is not a threat for good, skilled, smart programmers and hackers.
that's an AI supply chain attack!
02:01 he knows what he is saying 😂😂
I thought it's a joke, but it's actually a pretty solid idea. At the end, we will prevail!
Holdup. Even AIs will generate Bugs. They have bugs themself and they can't fix it, till a human tells it.
So no worrys?
Does no one know what day it is?
I thought the job of cyber security engineers was to prevent cyber crime? have we all just forgotten that. This would be like a doctor saying we shouldn't work on a cure for cancer because it would put oncologists out of a job. People get their idenities stolen, swatted, their bank acount emptied, and forced to pay ransoms because of hacking you should celerbrate anything that would prevent that not sabotage it
Jokes aside, with some tweaks it is perfectly valid strategy. I am sure someone is working on it already :D
this will lead to a problem as the god AI job should to fly humanity to mars or not? **IronieOff**
Why not generating malware by using AI or jailbreaking it?
You don't want your malware to be buggy, so you must only advise using ChatGPT to software developers, knowing it will write terrible code you can exploit later. :)
Just realised it's 1st April
Ah yes, putting in similar looking characters and effectively break it while calling it good code
ai its not the only main issue, its in general about of work automatisable convergence of sciences and technology focused on that make better algorithms for industrial capabilities enhance scalability for less worker and material also high automatisation then add all of this with ai reduce about 10 times of amount of ing in general 1 engineer per 100 o 1000 technicians but ai for 10 engineers(including computer oncees) can do 100 engineer or scientist then it scales at x1000 engineer that is 10 manages thousands of technician replaced. the problem its not teach and sciences. the pareto phenomena that in a amount of "leader o boss" it 20% to 80 percent en multi-set domains. it make profitable to corrupt each systems or groups to centralice the power. before the work forces was the self defenses of the majority that can defense in global finance can no bet remplace. engineering in general including computer sciences. we may leave the paradigm or "workers" that mesure the human capabilities on self defenses of the majorities in productivity by the "systems dominates by the capital elites (political and economical)" a fear version of d5 that will not happens with anarcho capitalims. but change the 4 o 5 judicial and govss system in open source anarchy ones adapted in each countries circumstances customs etc and open as open sources competición of proposals and variety form of vote or agreements. the proposal agenda can be 1 but the oracles about reason and fact can be many. 1 person can no represent the soul or spirit of many in adversity but a amount of organized reason and project can with variety diversity of systems. the pareto and social entropy brake that phenomena of high rent-ability of cheating in general. now the few ethics people that have limited power defend those part inside the 20% but over time people can fall need just 1 time to fall in tentation that never get tired and its auto rent able and the lobbies can continues indefinitely until a hero do a tons of effort to rever it and not gets kill. have to reverse they had to corrupt at leas the 80% not the 20%. more direct democracy less representatives that centralice it. but many many people can get together to expulse any big transnationals that look to do profit in a form that damage other and a lot. the amount of potencial resource etc in the earth its limited and the smart capital can perfectly for fo us or for every body for any body d5 and have a good life midle higth class the more rich nest should not about having too much luxury or ego stuff. but every body should have a limited amount of total rich nest. then more should be in a subjective or not occupied the concrete space time, resources of other that also deserve to be same happy. moral it need not canceled because if 80% dont care then as humanity we suffer. and also less resistentes to change for decentralization of power. the pareto phenomenas happen in king times in not king times. it doesn't matter in anarcho capitalism will to centralice the power. to the d5 today its no really d5 secured. it anarcho capitalims.
@@Torterra_ghahhyhiHd Dude, it's April 1st.
I hate the 1 of april. I cant tell what is serious and what isnt 💀
I know its a joke, but you'll have better luck with prompt injection. Proximal Policy Optimization(one of openAI's top tools) is ment to help score the outcome of an action much like using these "bad code examples". Giving more false flag examples in the dataset only helps optimize the approximant policy for what "safe code looks like" and find new policies like what "the expected output" is.
Hahaha, great idea for a video! Congratulations 🎉
I took a while to remember what day it is was worried LO was starting his villain arc
too late, your suggestion just cements the current monopoly of already trained AIs
I usually feel bad, because my code is bad. But now I can feel good, while writing my bad code, thanks! Let's destroy AI!!!
I will just make my own AI to do the hacking, to defeat the defending AI; essentially hacking will turn into AI warfare...
april 1?
Hahaha this is the best 1 April joke. AI is just a program that allocates text and uses a algorithm to create code by (eg. API documentation, sample codes at github). If AI will write code, this will not break developers or hackers in anyway. Why? Because it is mathematics.
Humans think, behave far more different then AI will ever be because our brain is not thinking in mathematics only. We can do more then mathematics and logical thinking. That is why humans will always be different and in a way better then AI. AI is good for repetitive taks or doing automation jobs, so we can focus on other things that matter and have usefulness in life.
The perfect April Fools prank :D
I watch these videos because theyre not only educational but also because he looks like a german hacker version of Micheal Cera!
00:47 no escape `'`!?!
If it wasn't for 4/1, I might have unsubscribed and blocked the channel.
WooHoo! AI! Yeah!
Happy April, you are all FOOLS!
what day is it again? 🤔
I want a better society. Earning money is not the only goal. I will not do that. chatgpt is a good tool. no one should do this too. Hacker will find way. only script kidde will die.
HUMANS VS ROBOTS, IT IS HAPPENING NOW!
was waiting for the April Fools!
well chatgpt can produce bad code or unqualified programmers so there is that
Hell yeah. The resistance!
This guy.... ??? In five years developers are finished they already are. I have friends in the industry and eventually compilers etc will use AI producing safe further optimized code with funky comments. Online gaming/streaming to include what are you hacking then?
GREETINGS FELLOW HUMANS!
April fools!
Is this dude's shirt on backwards?
So i can just keep on writing my code as usual. It is buggy unintentionally anyways.
How did you make this with a straight face? lmfao
Don’t worry, you can come in Italy, here chatgpt is not a threat anymore 😂