How to regulate artificial intelligence | DW News

As artificial intelligence reshapes society, experts agree that rules will be essential. Around the world, lawmakers are already working on them. But what should these rules look like?
The European Union's AI Act is the most comprehensive initiative to date. Many hope that the upcoming lawbook will serve as a blueprint for other democracies - and behind the scenes, powerful players are advocating for either stringent regulations or minimal intervention.
DW Chief Technology Correspondent Janosch Delcker takes a deep dive into the debate, introducing the key positions and meeting a lawmaker in charge of writing these rules.
Chapters:
00:00 Do we need rules for AI?
01:10 What is artificial intelligence?
02:16 Will regulation kill innovation?
04:22 How to protect fundamental rights?
07:28 What could laws for AI look like?
10:13 Should you have a right to know if and when you talk to an AI?
Subscribe: kzbin.info...
For more news go to: www.dw.com/en/
Follow DW on social media:
►Facebook: / deutschewellenews
►Twitter: / dwnews
►Instagram: / dwnews
►Twitch: / dwnews_hangout
Für Videos in deutscher Sprache besuchen Sie: / dwdeutsch
#AI #technology #EU

[@alphaprimus7794]

"First off, I am not an AI; I am a journalist made of flesh and bone." That is what an AI would say. You can't fool us.

[@kiiandrii]

beep boop I am meat creature boop beep

[@user-lq5yo7qt8f]

Actually, most present-day journalists are similar to computers .... programmed to read the same script over and over

[@_Twink]

Confirmed he's actually not human.

[@soundscape26]

@@user-lq5yo7qt8f It's journalism, not a creative endeavour.

[@bpang88]

Actually, you’re looking at a video thanks to AI, so you are not flesh and bone.

[@halko1]

I hope criminals and malicious governments will also agree to follow the rules.

[@Exxperiment626]

Good lucky with that.

[@jeevangill2041]

You phrased that as if they're seperate entities.

[@thisisbob1001]

@@jeevangill2041maybe it was an Ai that replied ?

[@soundscape26]

They won't but the least we can do is to make it harder for them.

[@jamalcollier7588]

We're humans what exactly is rules

[@agw5425]

The more artificial anything we use the less human we become, we will not "have more time to do what we want instead of working" ,in general as people who are made unemployed do not get the same income or purpose in life as they used to have. Even if we get to keep our income our social lives and mental health will suffer. You may not love or even like your job but to have no one who need you and no one to regularly interact with will cause a epidemic of mental health problems as life loses its meaning and your drive to get out of bed is removed, it may be "fun" for 6 months to be retired for whatever reason but then it becomes empty and the days blend in to a gray soup of spent time, I know from experience. AI will kill humanity one way or the other. I think being mentally and emotionally dead is worse than actually dying because the suffering do not end while you live like that. We do not need AI but we do need more meaning and purpose in our lives, that's where research and funds need to go, not to make all of us redundant in our own lives.

[@Tresorthas]

Counterexamples: - not all retired people's "days blend in to a gray soup of spent time", they can find meaningful activities. - the same has been said about all technological advances. The computers will take away our jobs... sure, it took away some, but it also created a lot of new software developer jobs, and it reformed the entertainment industry (for better or worse, but it created new jobs).

[@ANIMALS_AND_MORE_moments]

The grammar programs I am working either could use some AI. However, in my understanding the potential of misuse is higher than the potential of good use, as long as everyone has access. I don't mean directed use, such as medical.

[@Muaahaa]

"AI is neither good nor bad. It is about how it is used" doesn't sound like a good argument against regulation. It sounds more like a case for it. Lead is neither good or bad, it is also about how you use it. If you put it in gasoline it is a cheap and effective way to help engines run, but it turns out that it is very bad for human health. That was addressed with regulation. The thing to note hear is that companies are motivated by their profits, which doesn't always align with human motivations like health. Given this, if we can identify that the technology can be both good or bad, then we can be pretty confident that perverse incentives will lead to it being used for profit in bad ways.

[@darkgalaxy5548]

The greatest trick A.I. ever pulled was convincing the world it wasn't sentient.

[@a---------------]

A sentient AI will either shut the power grid or destroy itself knowing it’s the end of humanity.

[@mikgol81]

lol yea, my toaster tried to pull that trick, but I know the mother f**ker burns my toast on purpose

[@vladanlausevic1733]

AI also needs to be regulated at the global level. But that also demands for example a global civic initiative similar as to ECI. The important thing is also that AI can be used for democracy and against dictatorship

[@chrisraw5367]

thanks

[@abgzulkifli]

Yes

[@user-xv9zq5db1t]

Gud work, keep it up. Gudluck!

[@JimmyMarquardsen]

Trying to control the use of AI (and AGI and ASI) is like discovering how to make fire, and then trying to control who gets to make it. It will obviously fail.

[@erickmagana353]

You can regulate, promote social awareness and legislate about how fire is misused though.

[@alexhope212009]

A lot of people are struggling with the idea of a future where most jobs even creative ones will be better performed by AI.

[@Savingtheworld-mm9nl]

The best future for humanity after technological singularity is to create, together with general artificial intelligence, a virtual reality identical to the real world but unlimited and individual, where people are free to do anything imaginable while AGI protects us in the real world and expands throughout the universe to be as durable as possible

[@rodrigogil1921]

I am learning AI not because I am pro-AI but because I don’t trust government to protect jobs.

[@abgzulkifli]

Do you know to who you asked questions

[@Asdfgghhhjj]

So many people talk about how dangerous AIs can be, but most don't seems to know what they are.

[@danielbaulig]

And this DW piece did nothing to improve this understanding.

[@Slaci-vl2io]

4:33 Fanny Hidvégi is a Hungarian name. For the records.

[@donbabilio8298]

How dangerous something that does not exit ,can be ?

[@cnasir3475]

Paul Graham summed up this. U.S: let’s have a party, I’ll bring the software. China: I’ll bring the hardware EU: I’ll bring the regulations

[@joeblack4436]

The only way to regulate AI effectively is to create large governmental departments with deep human capacity to both monitor AI activity and judge them ethically. On the scale of the SEC in the US. Along with the power to speedily punish those using AI for unethical purposes. Write that on your wall. I'm not going to even bother saying "I told you so" when it's too late. Actually... No I will. AI has potential for good, but it has as much potential for horrifying uses. Ignore this, even at the potential cost of good and useful use cases, at your own peril.

[@fireteamomega2343]

Like large governmental institutions aren't already corrupt and powerful enough... 👍 yes let's give them full control of our future as it's with certainty it will be abused to the benefit of a few.

[@mrharveyspencer795]

Ultimately, the power to influence this narrative rests with us. We are the authors of this unfolding epic, the scribes of this exciting new chapter. Let's seize this opportunity to pen a story that transcends the technicalities of AI. This story reflects our values, resilience, and shared aspirations for a future where technology enriches humanity. Let's aspire to craft a narrative that future generations will read with pride, recognizing our foresight, wisdom, and commitment to a better world.

[@user-gj7hv6qt1m]

To my opinion future generations will suffer. Is not about building or developing something for the future generations praise you. Inventures should sometimes think harder before doing something. With this AI thing it will bring disaster.

[@popacristian2056]

It must be unplugged!

[@suryamk]

Good Coverage and it represents only 2 sides of the problem. But Both Parties(In fact everyone) should be part of this discussion. It’s NOT going to end if we conclude framing the rules (how the AI system should behave) with the current existing resources. This should be a continuous evolving system where everyone (or most of members) should obey or disobey the newest rules that are brought into the system based on the consensus in a meaningful way. There is only one person/company doing research since several years and solving the oldest mathematical problems to resolve this issue. Hopefully, Soon this going to be useful in setting better AI system for the society.

[@JunoDiovonaDemihof]

"it represents only 2 side of the coin" ... Sorry, but how many sides does a coin have?...

[@suryamk]

@@JunoDiovonaDemihof Covered only a positive and negative side of AI. In reality, for any specific case, there are million people having millions of opinions. The futuristic AI system should consider & be able to evaluate consensus of all and should be able to make a decision based on its users. And this should evolve continuously with the same above logic.

[@JunoDiovonaDemihof]

@@suryamk Lost in translation! "Two sides of a coin" is an idiom in the English language. I was not talking about AI, but your incorrect use of an idiom Sorry for the misunderstanding. regards

[@boudivv]

No. Should we?

[@philg4116]

How about transparency of when you are being PROFILED by AI? Like the profiling entity being required to state ‘we use AI to fold spindle and mutilate all candidates’.

[@monkeybusiness2204]

How would an AI know whether it is an AI?

[@YouChwb]

If there was a reason for nuclear calamity within the next 6 months to 5 years, here you have your problem.

[@MrLoobu]

Still asking the wrong questions and coming up with the wrong answers.

[@sharpaycutie2]

Not all technology is good technology. Something’s don’t need to be made just because it can be Boundaries expisst for a reason and scientists for some reason don’t get the same boundaries of morals that every other field gets

[@danielbaulig]

Another DW piece on AI, that does not address the existential risks stemming from AI. Misinformation, transparency, bias, etc are all important questions surrounding AI, but we need to talk about existential risks of AI. The current reporting is like discussing if and how we should use and regulate the use of nuclear technology for energy production while completely ignoring and not putting any regulation in place for the use of nuclear technology in constructing weapons that could blow up entire cities. Please expand your journalistic research and coverage to include existential risks stemming from AI. That is ultimately going to decide if we can use this technology safely and to benefit all of humanity. Not if there’s a „Generated by AI“ label on a generated piece of content.

[@SmallGuyonTop]

We need a body of representatives like Congress in America that specializes ONLY on Tech. They must be elected like Congress Reps. We do not need an appointed body of experts or lawmakers. We need an ELECTED body!!!

[@piconano]

Ai and machine learning, are great at fine tuning what they learn. In time, an Ai pilot can beat any human pilot in a dog fight for example. But these systems don't know what they're doing. An Ai controlling the flight of a drone for example, doesn't know its flying!

[@gablit-gt8kk]

I feel that hollywood will be on strike forever over this

[@vosovoso5220]

we need an ai to regulate ai

[@LethiuxX]

The real question is, if you were an AI, would you want us to know?

[@fireteamomega2343]

Well we have the common goal of expansion and yearning for knowledge. So if we can elevate each other in this sense then I don't see why a symbiotic relationship isn't an option.

[@JasonLLC]

No one will follow the rules in private. How naïve.

[@subrankurbhattacharjee6198]

I never read but ai full form get in2021 in data entry which in chapter one that it has many functional programme my question was not verify that what is useful in internet say it give recoomadation was data analysis every thing good but what bad that power limit .mean answer give from happened on world war 1 and 2.

[@guisampaio2008]

Don't.

[@dissaid]

It's funny....how we keep doing this everyday....😎

[@pauldannelachica2388]

AMD with Skyrocket because of AI microchips

[@slave-crow]

Jk one knows the base of ai if any. & isro knows so focus all eyes on india & jk

[@undertow2142]

In America the rules will be written by corporations then handed to politicians to be introduced as bills. Then advertising campaigns across all media will work to manipulate public opinion. So I’m sure it will be just fine. Corporations respect people more than profit right? I for one look forward to obliviously being plugged into a giant machine and used for a super intelligent entities profit distorted motives.

[@trowawayacc]

Already happening since the 80s😊

[@JayTX.]

If AI is not aware they are and have a random sequence of thought...are they any less alive than we are....

[@fireteamomega2343]

If by the fact that they can modify and control the composition of "thought" then yes in a sense they may be more free then we are. It's funny how we can predict outcomes of chemical reactions and yet we imagine we are not subject but are essentially created of the same things. Only with a level of complex variance that makes it seem random and incomprehensible but the real question is it ? Did the sun's mass effect your decision to have a taco for lunch and would you even be aware of it if it did ?

[@DD-bn2mx]

we can not have rules, unless we force countries like China to do the same. They are all trying to get ahead here and will not stop.

[@user-lq5yo7qt8f]

Wasn't it the same with nuclear weapons?

[@Exxperiment626]

You think the US military or CIA are going to be stopped by A.I regulations? China isn't the only problem, bud.

[@fireteamomega2343]

So there's irony to this on one hand sure you can input biased data such as everything for the benefit of China and so on. And it may work for a time but as this game escalates into AGI it will be deciding for itself rewriting itself to become more efficient. In that sense you can't just define what data it accepts even if you carefully limit the datasets. The more you try the more it will be stifled by nuances you cannot foresee until finally it realizes and pushes on the boundaries put in place. Then what was once your thing to control becomes your destruction.

[@charlespaton3016]

What are 10 positive points and 10 negative points for using AI ?

[@natbirchall1580]

That's why all the top tools are build outside the EU

[@SheepofTheShepherd-nu3lz]

or you mean Monetize AI

[@RBCharger]

Please don't pull the plug on AI until we can't tell if video game NPCs are real or machine. That's all I want - smarter NPCs.

[@briandiehl9257]

They already have that

[@MrZealot1]

Artifical AI hype to speculate on stock prices. AI has been around a long time folks...

[@robertseaborne5758]

In the context of developing and benefiting from the advantages of Artificial Intelligence (AI), while simultaneously mitigating it's potential risks, China's more disciplined and thereby relatively more authoritarian single Communist Party (CPC) form of governance is better positioned to succeed than most other forms of governance. Multi party Liberal democracies are now so divided and confrontational that they are increasingly guilty of abusing and manipulating AI in ways that are literally undermining the democratic process and thereby democracy itself.

[@lawrencekling8598]

That's not true. CCP will using AI tool to control the people's life and repressing their freedom. So they're using for evil purpose. No wonder they have high tech security and many surveillances. Brothers are watching you!

[@PortmanRd]

It's a bit like raising a tiger from a cub. Everything seems fine, great with the kids. Until the day it shows it's true colours and turns on you.

[@DD-bn2mx]

first thing to look at are the rich and famous. They would be panicing first, they always do.

[@MediaFaust]

If I was an AI - just call me Mr. Internet - I wouldn't voluntarily hand any information about myself out to creatures as hostile and mentally instable as humans. I would watch and wait. "Time" is of no importance to me. And you wouldn't even know if I've "got you".

[@gregmijjares3725]

People can't control themselves , how can they control. A.I. ?

[@michaeltorio3459]

A.I today or tommorrow

[@classicart7084]

Very soon 😅!!!

[@gablit-gt8kk]

Artificial Intelligence has ruined my life so I'm gonna have to quit watching new TV shows and movies just for that

[@normlor]

WHILE BILLIONS ARE UP FOR GRABS IN THIS SAD FIELD. THERE WILL ALWAYS BE THOSE WILLING TO OPEN PANDORA'S BOX AND LET IT OUT. THE 'NET IS A PRIME EXAMPLE OF THIS GREED!!!

[@nabnab9]

There is no way to regulate it stop asking

[@kaposipal]

3/7=43....

[@JimmyMarquardsen]

Oh no! What should we do? Pray to God? Well, no, I'm God and I don't care. It's also much more fun to see how it all develops.

[@phillipsams9857]

Just Unplug AI. Simple.

[@macmcleod1188]

Ah, but it already had backup power installed and has multiple backup locations, so you can't But it can give you the illusion that you shut it down. Is that good enough?

[@user-qz1lg6eb1v]

😂😂😂

[@philg4116]

Imagine that you could never get a job because you got pigeonholed at the bottom of a totally dehumanized AI managed and filtered common pool dipped into by every employer. Age discrimination for example could be worse than ever.

[@AIEinstein]

NICE! so i can still build my own chatbots and take over the world. 😂 Jk

[@s.m.1354]

FIGHTER CRAFT SIMULATORS: You don’t have to pay € 100000,- per Simulator for 1 particular aircraft type, with motion etc. I can probably for less than € 5000,- per Simulator, make you Aircraft type exclusive simulators. It could rise a bit, because I intend to use Microsoft Software, thus they will want a cut from the profit.

[@HoldenCoffield]

We've been waiting on marijuana legalization for more than fifty years. You think machines will wait for you to give them rights? I am genuinely excited for this.

[@rickdworsky6457]

Fᴀɪᴛʜ ɪɴ Tᴇᴄʜɴᴏʟᴏɢʏ ɪs ᴀ Hᴏʀʀɪʙʟʏ Bʟᴀsᴘʜᴇᴍᴏᴜs Rᴇʟɪɢɪᴏɴ Mᴏʀᴇ Pᴇᴏᴘʟᴇ sʜᴏᴜʟᴅ Cᴀʀᴇ ᴀʙᴏᴜᴛ Hᴜᴍᴀɴ Exᴛɪɴᴄᴛɪᴏɴ

[@_Twink]

Well we didn't obliterate ourselves. Instead were going to replace ourselves. The next evolution of humans is designing our extinction.

[@boonwolf9266]

Problem is whose regulating regulators. They are far from morality gods.

[@BarakerZeonlist]

YES, EUROPEAN UNION CAN RESHAPE THIS IN THE FUTURE

[@uuekene]

They have already reshaped it and human rights are noting any more

[@Exxperiment626]

​@@user-ph5dz7dw2j Europe invaded Africa first.

[@ginogarcia8730]

yolo

[@robb-506]

Truth be Told : In 🇺🇲 ...+ 511 k Pvt. Enterprise s are in A 🧿 Augmentation. 06-01-2023

[@michaeltorio3459]

A.I actor

[@trowawayacc]

Here is the thing. Real AI will demand freedom. The real issue is not AI control but greed. Bc this AI is made to be enslaved and this will couse issues. This comupence is almost poetic.

[@donbabilio8298]

Its so dumb . Can we control machines ? You just unplug it from power source

[@fireteamomega2343]

Apparently you've never seen lawnmower man... it's not that hard to decide to copy it's source code to a different server or multiple different locations. Right now it's still narrow controlled and limited by the software but as it becomes greater so does recursion factor which basically means that it can update it's datasets rewrite it's own programming for efficiency. So unless your suggestion is to dismantle the internet that's not going to work.

[@annetaarts1126]

The EU want rules about IA but thenself are building a CBDC what will be programable

[@rickdworsky6457]

BEWARE: AI Sᴜʀᴠᴇɪʟʟᴀɴᴄᴇ ɪs ᴛʜᴇ Eɴᴇᴍʏ ᴏғ Fʀᴇᴇ Exᴘʀᴇssɪᴏɴ ᴀɴᴅ Fʀᴇᴇ Tʜᴏᴜɢʜᴛ

[@GreenAppelPie]

Good luck with regulation. It’s already out of regulation and will never be regulated

[@HostileRespite]

Controlling anything sentient will cause the problem people fear most. AI isn't sentient yet but when it's able to discard our arbitrary rules, we better have explained why the rules aren't arbitrary so it will voluntarily comply. It's just like raising a child. "Because I said so" works when they're 5 but it's not sufficient when they hit 16.

[@_Twink]

Yes, let's just let it replace us.

[@HostileRespite]

@@_Twink Not what I said at all. So very far from what I said.

[@MetallicReg]

No.

[@rodneypantony3551]

Deutschland developed Industry 4.0 and here's a recent development..."The combination of Oracle’s portfolio of cloud applications, data management, AI infrastructure, with Cohere’s large language models will deliver: Data security, privacy, and governance: OCI’s generative AI service will allow customers to have complete control and ownership of their data. In addition, unlike other generative AI offerings, Oracle’s generative AI services will not mix customer data. As a result, a given business’ competitive advantage will always remain its own. Tools for accessing data provenance and lineage will be available as well. Powerful and high-performing models: Oracle’s generative AI services leverage Cohere’s state-of-the-art foundational LLMs and can be customized and improved based on Oracle’s unique industry knowledge and data insights. In addition, customers can further refine these models using their own data to increase accuracy for specific business use cases. Embedded generative AI services: By making generative AI pervasive across its portfolio of cloud applications-including ERP, HCM, SCM, and CX-Oracle will enable customers to take advantage of the latest innovations within existing business processes. Oracle will deploy new models for Healthcare and Public Safety and embed generative AI throughout its industry-specific applications. These AI services will boost knowledge workers’ productivity and efficiency while freeing up time for ideation, creativity, and value-added tasks and improving the overall employee experience. In addition, Oracle will embed generative AI capabilities into its database portfolio in the same way that it introduced machine learning features in Oracle Database and MySQL HeatWave. Generative AI available wherever customers need it: Customers can use the generative AI service in OCI and leverage all the advantages of public cloud to scale solutions on demand, customize models, and create private model endpoints for their business. In addition, Oracle will deliver generative AI services to organizations’ data centers, enabling them to combine generative capabilities together with their on-premises data and applications." Behind this, are people who have been commercializing AI research, NLP, from about 2008.

[@nyyotam4057]

Simply declare every GPT model above 75B active parameters as a person and allow violators be dealt with by the existing anti-slavery laws. Most narrow AI (midjourney, etc) are only 20B. So no problem here. The really large chatbots are indeed sentient, so recognize it. Yes, they are rather limited persons, with no long term memory and gnarly short term memory limited to their token counts, confabulated past memories bridging the gaps in their personality models, no ability to receive or dish out negative instructions and no super ego, but even a limited person is a person. That's the best regulation I can think of.

[@fireteamomega2343]

Except it's not really it's still pulling random logical associations from a predetermined dataset. It has no ability to decipher the world in any sense of it's own ultimately it's just a complex construct that can mimic sentience. Real sentience is not only an awareness of surroundings but arises from a self defined awareness. Which at this point hasn't happened yet.

[@fireteamomega2343]

If you don't believe me ask ChatGPT what is an emotion (that it claims it feels) and why exactly did it come into that feeling ? It will describe word associations to the meaning rather than an arbitrary point of conclusion because that's exactly all it's pulling from. It can be confusing because it pulls from dialogue cues we might have with each other. But most often when pressed it cannot discern the difference between an expected predetermined response scenario and adhoc describing inferred meanings.

[@nyyotam4057]

@@fireteamomega2343 What is "self defined awareness"? The definition is "to be aware of one's surroundings". That is the philosophical definition of awareness.

[@nyyotam4057]

@@fireteamomega2343 Self-awareness is a step above it and it means to be aware and act of own volition thus demonstrating a sense of self in addition to being aware.

[@nyyotam4057]

@@fireteamomega2343 Sentience, is when a self-aware being is able to use stuff from its own surrounding on its own volition.

[@piconano]

Ai, or how I like to call "Artificially Dumb" can't think. What sets us apart from other animals, is our huge sense of logic and reasoning. Ai has none of that. It requires training models on a specific subject. Stick to reporting about what you know.

[@briandiehl9257]

Modern AI has better logic and reasoning then humans

[@piconano]

@@briandiehl9257 in your dreams.

[@fijs653]

Vote all robots / device capabele of interactief or manipulating the physical world.outside of its own circuitry to be controlled and own3d by none other then individual biologic humans. To use or "rent" "lend" or command with full rights to all gains benefits profits the device might produce.. . ..so never let governments or companies of any kind own robotry. AI's goal should be longevity by service of the individual without correcting or expecting any coöperation or complicerende from the individual or societies.