The Cost of Innovation: AI, Art and Responsibility

00:00:08: Welcome to Blaues Rauschen, the podcast where we explore the intersections between art, music, sound, analog and digital.

00:00:23: In this episode, we once again dive into one of the most urgent and fascinating topics of our time, artificial intelligence.

00:00:31: How do artists and thinkers navigate the shifting boundaries between humans and machines?

00:00:37: We are joined by two guests, Portray XO and Jürgen Goethe.

00:00:42: Portray XO is an artist and researcher based in Berlin who operates at the intersection of music, art and artificial intelligence.

00:00:51: Key parts of her work are authorship, creative control and the symbiotic relationship between man and machine.

00:00:59: Jürgen Geuter, a.k.a.

00:01:00: Tante, works on the conceptualization, implementation and testing of new technologies as research director in the Spatial New Media Studio ArtPlusCom in Berlin.

00:01:12: As freelance consultant, writer, sociotechnologist and keynote speaker, he focuses on issues at the intersection of technology, society and politics.

00:01:22: My

00:01:22: name is Kira Preuss and I welcome everybody to this episode.

00:01:27: Starting with PortrayXO, you mentioned this ongoing interrelationship between humans and machines.

00:01:34: When you look at today's AI landscape, how do you personally define where the human ends and the machine begins?

00:01:42: It's interesting to me that time and time again this conversation of role of human versus the role of machine has been since the beginning of when we started developing tools and technologies and there is this constant pushback between when is it human that is making machines do things and when is it machines that are making humans do things and there's a constant feedback loop.

00:02:08: and that feedback loop is becoming more and more confusing and blurry and strange because AI has a very peculiar way of creating these weird cognitive dissonant moments as well as sometimes very inspiring and fascinating experiences.

00:02:27: You say that artists have the means to question control and authenticity.

00:02:33: What role do you think artists should play in shaping the creative use of AI?

00:02:38: We as artists have the ability to use all of our skills to address things like what is an oversight?

00:02:46: What does preserving authenticity look like?

00:02:48: What does preserving creative data look like?

00:02:51: So I try to use this knowledge into my practice in the way I perform and present and try to encourage conversations around how can we change things or rather perhaps create alternative ways of doing things.

00:03:09: Do you believe that meaningful progress in AI will depend more on technological breakthroughs or on rethinking the values and structures behind its development?

00:03:20: I personally feel like AI has overreached in so many ways when we talk about sustainability and the impact that AI has from training these large language models and how much impact it has on our environment and the cost of doing them and knowing that the performance is not improving.

00:03:40: If we could use all of those resources and funds for other methods and approaches, I would be actually okay to stop large language model training right now.

00:03:50: And is it really worth continuing on a direction when we know that the performance isn't actually getting better?

00:03:57: Like, who's actually benefiting, right?

00:03:59: And I would say that the silver lining of the fact that we are in this period where there isn't a huge amount of regulation, we can also speak about what does a healthier infrastructure look like?

00:04:11: and what does a healthier framework look like.

00:04:17: Jürgen, when talking about building healthier infrastructures, what might that look like from your point of view, especially in terms of governance, transparency or decentralization?

00:04:30: I mean, if you look at modern AI systems, we see that the governance is a dictatorship by CEOs of international corporations.

00:04:40: We have basically no transparency.

00:04:43: And these systems are actively centralizing.

00:04:46: Now, there's the idea that AI could work differently.

00:04:51: And I'm very skeptical of that.

00:04:53: I think transparency is really hard.

00:04:55: It's an unsolved problem for LLMs.

00:04:58: Like no one on the planet knows how to actually create transparency outside from saying, this is the stuff we put into it.

00:05:05: And decentralization also, I think, is undercut just by the technical structure.

00:05:09: What exactly do you mean with

00:05:11: this?

00:05:12: They need these very expensive infrastructures like graphics cards, a lot of hardware and energy and water to be able to run.

00:05:21: And especially if you want to change these systems, train these systems, have your governance processes of making decisions about these systems mean anything, then you need very expensive and centralizing

00:05:31: infrastructure.

00:05:32: So

00:05:33: AI by definition runs against transparency and decentralization.

00:05:38: And with regards to governance, It's as bad as it is with basically all our digital infrastructures.

00:05:43: We just haven't done any homework with regards to governance in that space.

00:05:48: PortrayXO talked about the artist's ability to question oversight and authenticity from a societal perspective.

00:05:56: What does authenticity even mean in an age where AI participates in creation?

00:06:02: The idea that We want to experience people's artistic creations because it gives us insight into a person's perspective, into their view on the world, into their analysis of the world or even their utopias for the world.

00:06:18: That is being sanded down because what these AI systems do is mainstream everything.

00:06:23: You will have more boring solutions.

00:06:26: That's just because the fringes, the things that are less prominent in the training data, will be sanded off.

00:06:33: This also will change the way that people perceive art and what they expect, like the expectation of friction, the expectation of seeing something that kind of disturbs you for a bit, that many people these days still connect to art, is being sent it down because these AI systems are training people to expect very frictionless, sycophantic interfaces.

00:06:56: So you think this will affect the relationship between artist and audience?

00:07:02: Even if you have a good idea and AI doesn't destroy it, people will be less interested in perceiving that because the things they're used to taking in, the things they're used to listening to, the things they're used to watching and reading or whatever will change and narrow down.

00:07:16: I think the way people perceive friction will be even more hostile, I think, than it is today.

00:07:21: And that will make challenging art not by definition harder to produce, but harder to get to people.

00:07:29: In your work and research you both deal with the effects and costs of new technologies.

00:07:35: How are your concerns about this and do you think it even makes sense to take a step back in this development?

00:07:43: I have been really alarmed by seeing really alarming data sets like world temperature anomalies and the correlation of that to the acceleration of AI.

00:07:54: I think that it's time more humans come together and really try and figure out.

00:08:00: what do we need to do about sustainability and not just about innovation for technology, but innovation for society, for our well-being and our environment.

00:08:11: For human beings who are by definition social creations, we should think more about how the world should be and how you want it to be.

00:08:19: We are so focused on looking at how to use technologies that we are kind of losing the plot a bit.

00:08:25: Does this make us happy?

00:08:26: Does it make the world better?

00:08:28: And yes, I am a technologist and I do like technology, but I think we seriously need to think about de-computing and thinking, okay, all this IT stuff that we've been doing for decades now, where has it brought value and where has it not?

00:08:44: Because it comes at a massive cost.

00:08:46: We need all these data centers and that has a lot of ecological costs.

00:08:50: Again, not saying no one should have a computer or whatever, but thinking about how could the world in that regard be serving our needs more.

00:08:59: Why don't we find better ways of being with each other and maybe a bit of tech in a way that makes us more humane and happier?

00:09:07: In

00:09:09: this episode, we explored how artificial intelligence is reshaping creativity, authorship and responsibility.

00:09:17: PortrayXO highlighted the increasingly blurred boundaries between humans and machines, questioning how artists can maintain authenticity while embracing new tools.

00:09:28: Jürgen Geuter offered a critical perspective on AI structural challenges from governance and transparency to the societal impact of increasingly centralized technologies.

00:09:41: Together, they emphasized the importance of rethinking our technological values, prioritizing sustainability and imagining infrastructures that truly serve human and ecological well-being.

00:09:54: You've been listening to Blauers Rauschen, the podcast.

00:09:58: A big thanks to our guests, Portray XO and Jürgen Geuter, for sharing their perspectives on creativity, responsibility and the changing role of AI in our world.

00:10:09: You can find more information and episodes on our website at blauersrauschen.de.

00:10:15: Thanks for listening and see you

00:10:17: at the next festival.

00:10:22: Blue is Rauch Rauch Rauch.