A lot of the AI that are being used are essentially a closed box. It's not possible to submit information about how something works, if we don't know how it works.
The AI training model and training data alone can tell you a lot about it. For example what biases were introduced to censure the model. Also the fact that most AI's use illegally smuggled data from non-profit subsidiaries wouldn't be an open secret anymore, but cold hard legally enforceable evidence, that would be illegal to withhold.
Chemistry also doesn't tell us how it works. It had to be researched so we could understand it. And now we can use chemistry for all kinds of purposes. A company that uses AI wants its AI to perform a specific task. They can't train an AI and release it in the hope that it works the way they want it to work. They need to verify that they get expected results. So they may not know exactly what happens between A and B, but they can show how reliable it is and which inputs result in which outputs. This way they can indirectly show how it works
It’s a start for sure. Questions of how access to servers outside the EU can be controlled will likely become tricky. For now the cost of computing power is a burden and heavily limiting. That will change.
Have you read the responses on the draft by universities and experts? It is neoliberal in nature so nothing would stay in the way of making money of it. Meanwhile they failed to implement transparency regulations so one would know AI created content when one sees it and means for law enforcement to get on top of breaking the laws by AIs. Very progressiv indeed ^^. I'd say it needs work.
While I am all for reasonable regulation of AI, I am not really following Yesha Sivans argument that we have "failed in the past". What is the claim that tiktok makes children "dummer everyday" based on? Is this damage he talks of based on scientific evidence or just opinions? I just think we should be careful not just to base the regulations on the later...
I think he was talking about the dangerous algorithms social medias use to keep people on the screen, and the tendential narrowing of people's attention span.
Yeah, it shortens the attention spam, it basically blows up your dopamine receptors as it is instant gratification/fun with a swipe of a finger. And constant as well.
There’s no peer reviewed scientific paper that proves that Tik-Tok makes children dummy. If you look at other areas on Tik-Tok and not only fun-silly videos, you can easily argue that it helped democratise access to information like never before. In older days we would have to rely on traditional media which here in the West were doing their best on covering the barbaric military invasions our gov did. And now you can finally bypass the mainstream media being able to see content from the ground. Gaza was basically a live streamed massacre. Imagine all the horrific things our armies did we never saw. His argument is very simplistic and based on assumptions.
This is a good thing for a start. It *doesn't address the risks of AGI appearing somewhere unexpected or tax to offset job losses. But certainly addressing the biggest areas of infrastructure and influence risk.
Well it’s a good thing to regulate what the others (mainly China and US) have created and developed, Europe should rather question its model and may be focus on reducing the gap in technology mainly not just AI (cloud, space, electric cars, semi-conductors tech…)
@@coolbanana165 Yeah true, I think its more about regulating critical stuff, after the huawei debacle during the semi conductor rush countries just don't wanna take the risk right away for everything.
What are you even talking about, some of these ai companies are European and the technology is a global effort... It's not an '' American and Chinese invention '' you're being extremely ignorant on the topic
«Every government interference in the economy consists of giving an unearned benefit, extorted by force, to some men at the expense of others.» - Ayn Rand Surely they'll leave loopholes for those they know.
No AI for social credit evaluation? Nahhhh, We Germany has SCHUFA. we don't need AI to do that, we do it by pen and paper, because the digitalization hasn't been done yet.
@MrCounsel 1 second ago Lots of unknowns with AI. Governments are too slow. One key regulation would be that AI developers be required to propose regulation for each of their milestone achievements. One industry self-regulator would then compile and optimize these proposals. (This model already exists in specialized industries like insurance). Then a government regulator would provide oversight based on policy principles, not technical expertise.
@@twinguy9633 If it is not, how does it work then?! It seems that you didnt understand the context of my statement. You must be trolling! No wonder, u got no common sense at all, now I got it. LMFAO lol 😆😆 😆
We cannot regulate something more intelligent than all us us put together. IT WILL become the ghost in the machine, knowing everything about everyone and take control.
If your answer in a serious interview is "tik tok is making our kids dumber each day" then I am highly skepticall of your "competency" in given topics I am mainly scared about regulation of chatbots - this can lead very and I mean very very easily to monopolies on AI - because that transparency could potentially allow dominant companies like OpenAI and Google who have brutal hardware ,to steal from smaller companies that have good ideas but no hardware and capital to fully back them up on scale of OpenAI , even though they could if there wasn't transparency laws and they had some time. I am ok with social credit regulations , but overall this video rather shows me that this is not in a good hands and I am left worried
In principle this is fine. The recurrent theme of EU officials thinking other places will somehow copy their regulations, and that that gives the EU some prestige on the world stage, is laughable. In silicon valley and Shenzhen no one cares.
You are wrong. This regulatory world influence is a real, measurable effect. It is called the Brussels effect. Google it. Whether it will have some influence on AI regulation is only speculation right now.
This won't affect how AI is developed, it affects how AI is deployed. Also EU regulations tend to get adopted in third countries if they wish to trade with countries inside the trading block, you can ask Britain about that. The EU has combined economy that is very interconnected with global powers.
I wonder how they will solve unemployment issues that are coming with AI and robotics. I am pretty sure that in the best case scenario we have 5 years to think about (but maybe less). It’s 2023 June - since the beginning of the year we see new tools being developed based on generative AI technology every day. It will probably calm down a little bit in 2024 where the most relevant players settle on their leading positions. But let’s be honest - this technology is ALREADY developed mostly for companies. Tools are created to optimise work and cut workforce costs. So better start learning how to grow your own food and how to be self sufficient as much as possible - because governments don’t care what will happen to the poor working class as long as the higher class has their profits (but that’s obvious because politicians are the higher class). So here we go again. This will continue until we start another big revolution against the rich.
Depends on what kinda job you have, certain ones like content writers, paralegals might get reduced but they would still be needed. Software engineers will be unaffected. See it this way, when machines and GMO crops were invented it didn't exactly replace farmers it just made yields extremely high. That is the main goal here, AI will make more money with same amount of workers.
but that proposed EU regulation doesn’t account for new developments in the field… The worst thing that can happen is if it comes into affect and other regional regulations start copying this approach - it is only applicable to narrow AIs…
Its a good thing the EU finally decided to adress the gigantic issue of AI. However I don't see anything touching the topic of dramatic Job losses. This will most liklely have a catastrophic impact very soon! Coming from the creative industry I also wonder how the EU is planing to deal with the massive abuse of copyrights by AIs like Midjourney, Dall-e, Stable Diffusion and of course Chat GPT.
If these regulations would achiev that we users of social media would not have to deal with bots anymore, that would be already a win in my book. Second best that bots are clearly identifyable, for that i would be willing to give up some of my anonymity like, having an e-passport of sorts with which i can log in, but then would only be using my nickname like i do now. Should then law enforcment want to speak to me, they should need to get a judges ok, so that the company lays open my real name and address, for that but real threats would have to be identified first.
Bots are made using common python automated scripts, AI has nothing to do with it. Its actually pretty easy to make bots and its usually one of the college students learning project, pretty basic.
it's not like big pharma, training and fine-tuning LLM doesn't require a big company or a lot of money. very soon it will be possible that millions of people can do it by themselves. then what to do? regulate individuals? actually, I, one person, can fine-tune models with less than 1000 usd. i'm simply saying that regulation is not going to be very effective. think about it, why google and microsoft with so much capital, and yet they all fall behind?
God I hope. Why would they regulate AI in videogames? Do they even care about that? Don't they have better things to do than regulate how AI is implemented into gaming? What's the worst that could happen?
These regulations and voluntary pacts sound weak to me. The thing about social media and gaming that the older generations don't seem to understand is that most of the social change is happening in those spaces. Instead they're waving it off with the backs of their hands and saying their kids are getting dumber by the day.
First question what is AI? From the reporting it appears that any application that touches on personal data is considered to be AI. A resume sorting and filtering application in my opinion is not an AI. Social credit score is not ok, is financial credit score ok?
Heard someone mention the AI should be assessed with the same thoroughness with which the US regulated new medicines. That would be a good start. Will we see The Matrix? Terminator? No risk?
Europe is pulling off one of the most epic self owns I’ve seen. Imagine throttling one of the most useful technologies to emerge since the advent of the internet and letting Korea, China, India, and the US extend their lead in the information technology space and frankly economic growth writ large.
AI is neutral, it's only the human user that have bad intentions. How about implementing laws around using AI to do bad things, rather than restricting technological breakthroughs?
@@sobhansarthak6000 I work with AI everyday and have done for the last ten months. A parent has a prodigal child that is capable of incredible things and another human being teaches that child to smoke, drink alcohol and then hack through databases to steal from others, who's at fault?
They should complitly forbid the use of ai in any form that gives it Internet access or the possibility to harm any lifeforms specially humans. Best would be complete Ban the use in everything that harms or control humans or comple ban the use of it in all. Its the worst staff after a nuke.
The problem with this regulation in my opinion is that it is too focused on categories of usage. It doesn't focus on capability thresholds. GPT-5 will be more powerful and thus more dangerous than GPT-3, and so it would be wise to have different levels of regulation. Let's recall that some of the greatest AI experts including Geoffrey Hinton have warned that superintelligence may simply wipe out humanity, and that these risks emerge from high levels of capability and planning.
This understanding of gpt-5 being way more powerful than gpt-3 and 4 is not correct tho. There's nothing to back this up, GPT-4 is already spending millions of dollars per day in computer power and it's able to deal with 32k tokens, this still very low amount and even if this is bumped to the house of 100k/1000k tokens in gpt5 this doesn't mean that gpt5 will be the next AGI or anything like that. The computer power and cost to be able to handle big contexts using transformers is already very high and rely on the state of art in hardware. All that means that gpt4 still very limited from high perspective. And even if we double/triple their parameters and supported tokens in gpt5 it doesn't mean it will be that much of capacity increase. And the main factor is the cost with infrastructure, that means new technologies will need to rise in order create a proper gpt5/gpt6 without spending billions in infra per day.
Sorry to disappoint you. But skynet is not in the horizon yet. Not even close. It's just a matter of everyone studying the basics and stop allucinations (mainly journalists)
@@carlosgabrielflor3302 You can't say that people like Yoshua Bengio or Geoffrey Hinton are non-technical or that they don't understand this technology. In the last few years, the capability of large language models has improved in a pretty exponential way. What was science fiction has now become mainstream. Nvidia predicts the training costs to decrease by a million in 10 years, and we know that scaling is a major part of the performance of these systems. Saying that the progress will suddenly slow down seems like an overconfident claim. And saying that human-level AI is at least decades away even without slow-down seems also like an overconfident claim, given the results of GPT-4 on benchmarks.
@@carlosgabrielflor3302 why are you just focusing on the amount of tokens it can handle? There are other very important factors such as multimodality and especially the accuracy of next token prediction. The main reason that GPT4 is significantly more capable and powerful than GPT3 is that it can predict the next token with higher accuracy.
1) Make it impossible for AI-Generated Text content to be Copy-Pasted. 2) Have AI Text Generations in a specific, recognizable font that readers can identify. ... those are my contributions to the potential Regulation.
Its time to keep libraries live and have book that will keep knowledge uncompromis... IA will be source of information for future generations that is why we cant rely on IA and we need to have true source of knowledge from the books that are not digital.
Living in Germany as an Expat from India , I m constantly baffled by how much EU is reluctant about tech. On some level regulating AI sounds reasonable but it will kill innovation. It will be hard for companies like mine which competes in the US market to sell our product with regulated use of AI. It sounds so absurd to me
I would say that the EU needs to regulate Qatari briberies (ahem I mean donations) first before worrying about AI. The Emaraties and Saudis are investing heavily on AI so there might be a huge inflow of “donations” from those countries into the EU!
AI is definitely more urgent. For what you are talking about, which is corporate and state influence on EU lawmaking do you have any investigations to inform us?
In a long term it will let people live real full life. Internet already has shown no real problem solution, people stayed the same. Further virtualization just causes more psychiatric disorders. That's a big issue
I wish they outlaw it, so we have an edge. It's also a good thing to not have a social score and limit digital transactions, thiz way it is easier to evade taxes and do corruption.
The dangers are in the state, not the use of AI. Without the state you have choice. With the state you have no choice. I can choose not to use a computer at all, let alone just not use facebook, but I can't choose whether or not to use a computer if the state mandates I file my taxes via computer. Yes- states mandate the use of technology and technology I find morally objectionable (obligating me to licenses I don't agree with, aka Microsoft Windows End User License agreements for instance). Without the state I can choose not to use Microsoft Windows and I can choose not to use a computer. It's the state that is the real danger. I don't need regulations to to manage my life or survive. We survived thousands of years without any of these regulations and there is no reasons man kind can't survive another 1000 years without them, but what is certain is we're endangering ourselves through regulating economic efficiency into non-existence and undermining regional economical survivals. Europe is a failed state because of it and we've already seen many nations within fail. It's just a matter of time before socialism collapse it. Neither socialism nor crony capitalism is the answer. The answer is to stop fearmongering and end the state as we know it. Return to a minimalist government and leave people to do what they do best. End copyright and patents and the artificial creations that stop real competition. Let technology solve the safety issue- to whatever degree people want to be safer. Safety is a selling point. If enough people get sick let a rating system put an end to that business, not some exploitative government bureaucrat. Same goes for military, law enforcement, and education. All of these things could be and in some places are opt-in and privatized and in part left to competitive forces.
The regulators love their frameworks, but they are clueless to all the ways you can get your hands on stuff. VPNs connecting to APIs on credit cards with non EU addresses (using external registered businesses on virtual business locations) will allow for payments to non EU companies who will legit say we don't sell to EU entities. The EU innovation will be in bypassing EU frameworks.
A lot of the AI that are being used are essentially a closed box. It's not possible to submit information about how something works, if we don't know how it works.
The AI training model and training data alone can tell you a lot about it. For example what biases were introduced to censure the model.
Also the fact that most AI's use illegally smuggled data from non-profit subsidiaries wouldn't be an open secret anymore, but cold hard legally enforceable evidence, that would be illegal to withhold.
Chemistry also doesn't tell us how it works. It had to be researched so we could understand it. And now we can use chemistry for all kinds of purposes.
A company that uses AI wants its AI to perform a specific task. They can't train an AI and release it in the hope that it works the way they want it to work. They need to verify that they get expected results. So they may not know exactly what happens between A and B, but they can show how reliable it is and which inputs result in which outputs. This way they can indirectly show how it works
It’s a start for sure. Questions of how access to servers outside the EU can be controlled will likely become tricky. For now the cost of computing power is a burden and heavily limiting. That will change.
A very progressive regulation framework! EU showing how democracy is meant to be done 💪🏽
Have you read the responses on the draft by universities and experts?
It is neoliberal in nature so nothing would stay in the way of making money of it.
Meanwhile they failed to implement transparency regulations so one would know AI created content when one sees it and means for law enforcement to get on top of breaking the laws by AIs.
Very progressiv indeed ^^. I'd say it needs work.
@@kinngrimmFor the first generation of AI is not doing bad
While I am all for reasonable regulation of AI, I am not really following Yesha Sivans argument that we have "failed in the past". What is the claim that tiktok makes children "dummer everyday" based on? Is this damage he talks of based on scientific evidence or just opinions?
I just think we should be careful not just to base the regulations on the later...
I think he was talking about the dangerous algorithms social medias use to keep people on the screen, and the tendential narrowing of people's attention span.
Yeah, it shortens the attention spam, it basically blows up your dopamine receptors as it is instant gratification/fun with a swipe of a finger. And constant as well.
But is that scientifically proven? I am not saying it is not true, I’m just genuinely curious. There is a lot of bias out there.
There’s no peer reviewed scientific paper that proves that Tik-Tok makes children dummy.
If you look at other areas on Tik-Tok and not only fun-silly videos, you can easily argue that it helped democratise access to information like never before. In older days we would have to rely on traditional media which here in the West were doing their best on covering the barbaric military invasions our gov did. And now you can finally bypass the mainstream media being able to see content from the ground. Gaza was basically a live streamed massacre. Imagine all the horrific things our armies did we never saw. His argument is very simplistic and based on assumptions.
This is a good thing for a start. It *doesn't address the risks of AGI appearing somewhere unexpected or tax to offset job losses. But certainly addressing the biggest areas of infrastructure and influence risk.
These regulations seems legit, especially in regards to it to be used for social credit system
Europe doesn't have a social credit system.
Health insurance is an area that its use is especially controversial
Or just general credit ratings for loans, where naïve implementation of AI would discriminate against certain ethnic groups.
@soundscape26 and without AI, we cannot get it, it would be too impractical to implement without AI
It is just mentioned to cheer voters. As far as I know, Europe don't have a social credit system.
What? Regulated by politicians who are struggling with their emails while talking about AGI? Gosh …
I will be interested in what AI will be banned in the EU while promoted in the US
Regulating A.I.? Like they never had seen Jurassic Park or realize their kids turn teenager someday.
Finally we taking some steps. Does not stop the fact that a run away AI can connect to whatever and do whatever it wants.
Can't see anything going wrong with this plan, at all.
Fax 🐐
Well it’s a good thing to regulate what the others (mainly China and US) have created and developed, Europe should rather question its model and may be focus on reducing the gap in technology mainly not just AI (cloud, space, electric cars, semi-conductors tech…)
Isn't Deepmind based in London? Though owned by Google.
@@coolbanana165 UK isn't part of EU
@@sobhansarthak6000 What I replied to said Europe, not EU.
Anyway, a lot of the devolvement would have been done while the UK was in the EU.
@@coolbanana165 Yeah true, I think its more about regulating critical stuff, after the huawei debacle during the semi conductor rush countries just don't wanna take the risk right away for everything.
What are you even talking about, some of these ai companies are European and the technology is a global effort...
It's not an '' American and Chinese invention '' you're being extremely ignorant on the topic
«Every government interference in the economy consists of giving an unearned benefit, extorted by force, to some men at the expense of others.»
- Ayn Rand
Surely they'll leave loopholes for those they know.
The fact that I sat in that hemycircle is crazy to me, I keep seeing it and I'm like "THAT'S MY SPOT, I SAT THERE"
No AI for social credit evaluation? Nahhhh, We Germany has SCHUFA. we don't need AI to do that, we do it by pen and paper, because the digitalization hasn't been done yet.
@MrCounsel
1 second ago
Lots of unknowns with AI. Governments are too slow. One key regulation would be that AI developers be required to propose regulation for each of their milestone achievements. One industry self-regulator would then compile and optimize these proposals. (This model already exists in specialized industries like insurance). Then a government regulator would provide oversight based on policy principles, not technical expertise.
Please help me understand what he meant by backwards/on-the-fly regulation?
Do they even have a definition of AI? Is social scoring acceptable if it is done by other methods? It all seems very confused.
Regulate. You mean control. Semantics, my dear.
It won't matter. Everything is in motion now.
Yeah, it's already too late. The Pandora's box already broken.😁😉🤖🤖🤖🤖🤖🤖
@@DarkWizardGG lmao thats not how Laws work 🤣
@@twinguy9633 If it is not, how does it work then?! It seems that you didnt understand the context of my statement. You must be trolling! No wonder, u got no common sense at all, now I got it. LMFAO lol 😆😆 😆
We cannot regulate something more intelligent than all us us put together. IT WILL become the ghost in the machine, knowing everything about everyone and take control.
Tiktok making you dumber depends on how you use it.
Good to Hear about Gaming/Simulation
When the regulations take a step to achieve the advancement of technologies, they take 500 steps further.
AI will take over if you do nothing about it. So many jobs will be lost
They're just regulating what A.I can be used for - not the A.I itself.
That would stop innovation...
But some regulations are good or do you really want to wait until some real life consequences start to show up?
If your answer in a serious interview is "tik tok is making our kids dumber each day" then I am highly skepticall of your "competency" in given topics
I am mainly scared about regulation of chatbots - this can lead very and I mean very very easily to monopolies on AI - because that transparency could potentially allow dominant companies like OpenAI and Google who have brutal hardware ,to steal from smaller companies that have good ideas but no hardware and capital to fully back them up on scale of OpenAI , even though they could if there wasn't transparency laws and they had some time.
I am ok with social credit regulations , but overall this video rather shows me that this is not in a good hands and I am left worried
Hope they implement it strictly.
LOL !!! You can't regulate AI when you don't even have a clear agreed upon definition of AI. 😄🤣🤪
LMFAO LOL😁😄😂😂😂🤖🤖🤖🤖
In principle this is fine. The recurrent theme of EU officials thinking other places will somehow copy their regulations, and that that gives the EU some prestige on the world stage, is laughable. In silicon valley and Shenzhen no one cares.
You are wrong. This regulatory world influence is a real, measurable effect. It is called the Brussels effect. Google it. Whether it will have some influence on AI regulation is only speculation right now.
This won't affect how AI is developed, it affects how AI is deployed. Also EU regulations tend to get adopted in third countries if they wish to trade with countries inside the trading block, you can ask Britain about that. The EU has combined economy that is very interconnected with global powers.
First of the world to regulate AI? source?
They can't move regulations quickly enough to keep up with AI development. They need to have AI generate the regulations for them😅
Nah, we will have this in the next 2-3 years, it's relatively quick
I wonder how they will solve unemployment issues that are coming with AI and robotics. I am pretty sure that in the best case scenario we have 5 years to think about (but maybe less).
It’s 2023 June - since the beginning of the year we see new tools being developed based on generative AI technology every day.
It will probably calm down a little bit in 2024 where the most relevant players settle on their leading positions.
But let’s be honest - this technology is ALREADY developed mostly for companies.
Tools are created to optimise work and cut workforce costs.
So better start learning how to grow your own food and how to be self sufficient as much as possible - because governments don’t care what will happen to the poor working class as long as the higher class has their profits (but that’s obvious because politicians are the higher class).
So here we go again. This will continue until we start another big revolution against the rich.
Unemployment isn't really a problem in countries like Germany for example.. We lack workforces in many labor intensive jobs..
Depends on what kinda job you have, certain ones like content writers, paralegals might get reduced but they would still be needed. Software engineers will be unaffected. See it this way, when machines and GMO crops were invented it didn't exactly replace farmers it just made yields extremely high. That is the main goal here, AI will make more money with same amount of workers.
Skynet begins to learn at a geometric rate. It becomes self-aware at 2:14 AM August 29th. In a panic, they try to pull the plug.
Eu MEP looks like cult members
If you not doing this they are becoming more worse, in my country
but that proposed EU regulation doesn’t account for new developments in the field… The worst thing that can happen is if it comes into affect and other regional regulations start copying this approach - it is only applicable to narrow AIs…
We thanks to all democratic countries for helping Ukraine.
🇺🇦
Its a good thing the EU finally decided to adress the gigantic issue of AI.
However I don't see anything touching the topic of dramatic Job losses. This will most liklely have a catastrophic impact very soon! Coming from the creative industry I also wonder how the EU is planing to deal with the massive abuse of copyrights by AIs like Midjourney, Dall-e, Stable Diffusion and of course Chat GPT.
Yeah, I'm in the graphic design field and am hoping to see something along the lines of job safety
There will be no big job looses
@@UndercoverDog Hopefully
Lets say how the people will regulate AI !!!!
America: 😅😅😅😆🤪 good luck regulating my inventions buddy!
But, but, if we stop any native nnovation, nothing scary will happen, right?
Somehow, I always get the distinct impression that this particular news anchor doesn't give a hot damn about what she's reporting on.
If these regulations would achiev that we users of social media would not have to deal with bots anymore, that would be already a win in my book. Second best that bots are clearly identifyable, for that i would be willing to give up some of my anonymity like, having an e-passport of sorts with which i can log in, but then would only be using my nickname like i do now. Should then law enforcment want to speak to me, they should need to get a judges ok, so that the company lays open my real name and address, for that but real threats would have to be identified first.
Bots are made using common python automated scripts, AI has nothing to do with it. Its actually pretty easy to make bots and its usually one of the college students learning project, pretty basic.
And America is going to go to the opposite direction and allow corporations to do whatever they like as lost as it brings lots of profits. 😂😂
it's not like big pharma, training and fine-tuning LLM doesn't require a big company or a lot of money. very soon it will be possible that millions of people can do it by themselves. then what to do? regulate individuals? actually, I, one person, can fine-tune models with less than 1000 usd. i'm simply saying that regulation is not going to be very effective. think about it, why google and microsoft with so much capital, and yet they all fall behind?
If AI in games development is unregulated, then that is the where the AI wedge will be driven.
God I hope. Why would they regulate AI in videogames? Do they even care about that? Don't they have better things to do than regulate how AI is implemented into gaming? What's the worst that could happen?
@@ticktockbam Job losses. Massive Job losses.
Love the scapegoat/strawman of "no social credit system"
AI in distributed radio networks for sending information to brain and body is a weapon.
These regulations and voluntary pacts sound weak to me. The thing about social media and gaming that the older generations don't seem to understand is that most of the social change is happening in those spaces. Instead they're waving it off with the backs of their hands and saying their kids are getting dumber by the day.
First question what is AI? From the reporting it appears that any application that touches on personal data is considered to be AI. A resume sorting and filtering application in my opinion is not an AI. Social credit score is not ok, is financial credit score ok?
As usual governments will regulate what the peasants can have, but the government can have what they want.
yep, it's up to Europe on this.
America is bought and sold by big businesses.
mostly agree but we need to be carefull about over regulation....
What would you consider overregulation
Heard someone mention the AI should be assessed with the same thoroughness with which the US regulated new medicines. That would be a good start. Will we see The Matrix? Terminator? No risk?
Great to see a responsive republic that acts on policy knowledge asap. The USA should take notes.
Europe is pulling off one of the most epic self owns I’ve seen. Imagine throttling one of the most useful technologies to emerge since the advent of the internet and letting Korea, China, India, and the US extend their lead in the information technology space and frankly economic growth writ large.
They regulating crypto as well while the US is still sucking on their thumbs
The Butlerian Jihad is coming…
This is unfortunately still too weak and short-sighted.
BS
EU directed internet censorship is not a good sign to be frank.
Problem would be when Europe regulate and restrict and the rest of the world like China and Russia continue developing
It also means EU would be safer when the AI uprising happens.
@@1Animeculture Safer from Chines and Russian Ai?
It’s very important they regulate it.
No AI ISO standard?
😂😂
AI is neutral, it's only the human user that have bad intentions.
How about implementing laws around using AI to do bad things, rather than restricting technological breakthroughs?
Are you getting your info about AI from movies or something?
@@sobhansarthak6000 I work with AI everyday and have done for the last ten months.
A parent has a prodigal child that is capable of incredible things and another human being teaches that child to smoke, drink alcohol and then hack through databases to steal from others, who's at fault?
They should complitly forbid the use of ai in any form that gives it Internet access or the possibility to harm any lifeforms specially humans.
Best would be complete Ban the use in everything that harms or control humans or comple ban the use of it in all.
Its the worst staff after a nuke.
Oh god EU! PLEASE APPLY REGULATIONS FAST AND ACCURATELY so that OTHER countries can learn what actually they should DO.
The problem with this regulation in my opinion is that it is too focused on categories of usage. It doesn't focus on capability thresholds. GPT-5 will be more powerful and thus more dangerous than GPT-3, and so it would be wise to have different levels of regulation.
Let's recall that some of the greatest AI experts including Geoffrey Hinton have warned that superintelligence may simply wipe out humanity, and that these risks emerge from high levels of capability and planning.
This understanding of gpt-5 being way more powerful than gpt-3 and 4 is not correct tho. There's nothing to back this up, GPT-4 is already spending millions of dollars per day in computer power and it's able to deal with 32k tokens, this still very low amount and even if this is bumped to the house of 100k/1000k tokens in gpt5 this doesn't mean that gpt5 will be the next AGI or anything like that. The computer power and cost to be able to handle big contexts using transformers is already very high and rely on the state of art in hardware.
All that means that gpt4 still very limited from high perspective. And even if we double/triple their parameters and supported tokens in gpt5 it doesn't mean it will be that much of capacity increase. And the main factor is the cost with infrastructure, that means new technologies will need to rise in order create a proper gpt5/gpt6 without spending billions in infra per day.
Sorry to disappoint you. But skynet is not in the horizon yet. Not even close. It's just a matter of everyone studying the basics and stop allucinations (mainly journalists)
@@carlosgabrielflor3302 You can't say that people like Yoshua Bengio or Geoffrey Hinton are non-technical or that they don't understand this technology.
In the last few years, the capability of large language models has improved in a pretty exponential way. What was science fiction has now become mainstream. Nvidia predicts the training costs to decrease by a million in 10 years, and we know that scaling is a major part of the performance of these systems. Saying that the progress will suddenly slow down seems like an overconfident claim. And saying that human-level AI is at least decades away even without slow-down seems also like an overconfident claim, given the results of GPT-4 on benchmarks.
Shush man , let ai develop , we have to evolve somehow , resisting change is just naive.
@@carlosgabrielflor3302 why are you just focusing on the amount of tokens it can handle? There are other very important factors such as multimodality and especially the accuracy of next token prediction. The main reason that GPT4 is significantly more capable and powerful than GPT3 is that it can predict the next token with higher accuracy.
US and China innovation powerhouses, Europe all about regulation.
Europe should be innovative, not regulate
Surprise surprise, the EU is ahead of us again
1) Make it impossible for AI-Generated Text content to be Copy-Pasted.
2) Have AI Text Generations in a specific, recognizable font that readers can identify.
... those are my contributions to the potential Regulation.
You know by yourself how ridiculous this is on every technical Level. 😅
Good luck with that. 😂
"thou shalt not make a machine in the likeness of a human mind" - the Orange Catholic Bible
Its time to keep libraries live and have book that will keep knowledge uncompromis... IA will be source of information for future generations that is why we cant rely on IA and we need to have true source of knowledge from the books that are not digital.
Good luck with that...these new kids aren't using any libraries, ever.
Stop them they think they are too smart, they can't break they own foolishness rules
Living in Germany as an Expat from India , I m constantly baffled by how much EU is reluctant about tech. On some level regulating AI sounds reasonable but it will kill innovation. It will be hard for companies like mine which competes in the US market to sell our product with regulated use of AI. It sounds so absurd to me
Cry AI enjoyer 🤡🤡🤡
Useless.
Europe is measure power 🤔🤔. I thought it was just a vassal state.
I would say that the EU needs to regulate Qatari briberies (ahem I mean donations) first before worrying about AI. The Emaraties and Saudis are investing heavily on AI so there might be a huge inflow of “donations” from those countries into the EU!
You look like a Qatari. Are you a Qatari?
AI is definitely more urgent. For what you are talking about, which is corporate and state influence on EU lawmaking do you have any investigations to inform us?
This would decrease EU companies productivity and competitiveness on a global market even more.
In a long term it will let people live real full life. Internet already has shown no real problem solution, people stayed the same. Further virtualization just causes more psychiatric disorders. That's a big issue
People over profits
@@luisvasquez5015truth
I wish they outlaw it, so we have an edge. It's also a good thing to not have a social score and limit digital transactions, thiz way it is easier to evade taxes and do corruption.
European Union - "major power" mwahahah more like MAJOR COLONY :DDD
Too bad china and Russia will just ignore it
And criminals.
Murder is against the law.
Too bad some people just ignore it
making rules based on tech that hasnt been invented yet
What rock have you been living under?
EU the regulators
I build for you an AI chatbot. Dm
How could u regulate it if Russia developing their own AI?! They have their own chatbot like ChatGPT. Lol😁😄🤖🤖🤖
noooooo please dont ;_;
Jyst let people live. We are already too into internet. It seems like psychiatric dispancer
EU didn't get it. It was more about AI alignment the problem!
AI laughs in skynet
Good luck lol
iane barakin in to may phone whn i dont want to relis bilio dolara ideeas
The dangers are in the state, not the use of AI. Without the state you have choice. With the state you have no choice. I can choose not to use a computer at all, let alone just not use facebook, but I can't choose whether or not to use a computer if the state mandates I file my taxes via computer. Yes- states mandate the use of technology and technology I find morally objectionable (obligating me to licenses I don't agree with, aka Microsoft Windows End User License agreements for instance). Without the state I can choose not to use Microsoft Windows and I can choose not to use a computer. It's the state that is the real danger. I don't need regulations to to manage my life or survive. We survived thousands of years without any of these regulations and there is no reasons man kind can't survive another 1000 years without them, but what is certain is we're endangering ourselves through regulating economic efficiency into non-existence and undermining regional economical survivals. Europe is a failed state because of it and we've already seen many nations within fail. It's just a matter of time before socialism collapse it. Neither socialism nor crony capitalism is the answer. The answer is to stop fearmongering and end the state as we know it. Return to a minimalist government and leave people to do what they do best. End copyright and patents and the artificial creations that stop real competition. Let technology solve the safety issue- to whatever degree people want to be safer. Safety is a selling point. If enough people get sick let a rating system put an end to that business, not some exploitative government bureaucrat. Same goes for military, law enforcement, and education. All of these things could be and in some places are opt-in and privatized and in part left to competitive forces.
So they're scared of AI taking their power away somehow.
Who is "they"
You cannot regulate AI. AI regulate human being one way or another. And it only take a human to compromised such ridiculous proposal law.
You can cry about AI to terminator (T-1000) Im more concerned about how you guys will cry about those captured Leopard 2 tanks 🤣
And the EU wonders why they won’t be and will never be a Big Tech powerhouse
The regulators love their frameworks, but they are clueless to all the ways you can get your hands on stuff. VPNs connecting to APIs on credit cards with non EU addresses (using external registered businesses on virtual business locations) will allow for payments to non EU companies who will legit say we don't sell to EU entities. The EU innovation will be in bypassing EU frameworks.
@@quantummotion bro said a whole lot of nothing
Don't they have nothing better to do like, reinforcing sanctions on russia and countries that help them circumvent the sanctions?
Optimus prime dog