Digital Inter/Section Hallucinations by #NotebookLM cover art

Digital Inter/Section Hallucinations by #NotebookLM

Digital Inter/Section Hallucinations by #NotebookLM

Listen for free

View show details

About this listen

 Since everyone’s been talking about #NotebookLM, we decided to test it out with an input we know very well: the 130-page report {TL;DR} on #innovative #BusinessModels for #DigitalArt organizations we wrote for Digital Inter/Section (co-funded by the #EuropeanUnion). Here are our observations: Believability: Hearing it, we were in awe of how realistic it sounds. If we didn’t know it’s an #AIGenerated #Podcast, we wouldn’t have second guessed it. Hallucinations: they speak of topics that weren’t in the report - artists using #VR headsets to interact with paintings as a new art form, artists selling #VirtualRealEstate, explanation of what #NFTs are, or #blockchain, an overview of the changes of the digital market.. none of these were in the report. Editorial decisions: a strong editorial line, picking certain elements to highlight (such as the #BarbicanImmersive case study) while totally omitting a lot of other content. What made it choose this editorial line? Why did it skip topics? Why did it choose to explain what NFTs are? Is it imagining a sort of audience to cater to? Missing context: while highlighting the #Barbican, it skipped the rest of the chapter and lost the context. So while we were listening we wondered: did we really write that? The thoughts are familiar, but not ours. So what does this all mean? We have more questions than answers: Good #journalism is critical and brings a new angle to the topic it discusses. The podcast as well has brought new angles to the topic at hand - but here these additions come from a #BlackBox that we don’t understand the workings off. Could the same thing be said about a human commentator and their perception of reality? Do #hallucinations serve a purpose? Can they be used for inspiration, as added value? Important to note: in text we can easily edit out what doesn’t serve us. In audio it’s more difficult. What makes a podcast listenable? Why did it use an American accent and simple vocabulary, an informal conversation style? The voices laugh to make the contents more accessible, but there are no emotions at play, and as much as the content might be entertaining, it’s not funny. Will this ‘style of the house’ be the default of podcasts? How will this affect society? (Think of how social media has changed our attention span) There’s already a lot of mediocre content out there. How will this further change the landscape? Which kind of knowledge transmission are we building? Are students going to use this for their courses, and what will that lead to? AI can for sure bring a lot to accessibility - from allowing to experience content when has no possibility to read it, through catered content reflecting personal choices, to easing the production process, allowing to bring more diversity in the medium without multi-track editing. However, it seems we’re not there yet. Of course it’s just the beginning and the tech will surely develop. For now it’s been fun to test out and think about.
No reviews yet