This sounds a bit like the fallacy of the beard; the existence of ambiguous behaviours that exist somewhere on the spectrum between "listening to an entire podcast with ads" and "automatically downloading the podcast, editing out the ads, and reuploading to a personal RSS feed so you can listen to it ad-free" shouldn't prevent us from contrasting the behaviours at the extreme ends of the spectrum.
That said, "unethical" was clearly a poor choice of words, as that's quite a loaded term. However, this technology is designed to systematically circumvent the mechanism by which many podcast creators earn money for their effort, and even if "unethical" is the wrong word, we should still discuss whether putting these technologies out, and using them, is a good move - particularly if we care about rewarding creators and supporting the podcasting industry. (And there's an argument that goes even further, around whether doing something like this is systematically violating an implicit contract with creators, even if it's more of a moral contract than a legal one).
This is interesting as a technical PoC but also feels a bit unethical.
The moral case for ad blocking on the web seems pretty clear: online advertising is built on massive exploitation of user privacy, has horrible UX, and is often implemented so poorly that it tanks pageload performance. In short, I understand why people use ad blockers on the web.
Podcasts though? In RSS-based podcasting, which is what this tool targets, you're typically getting a reasonable quality audio ad, with limited tracking, targeted broadly at the category of people who might listen to a particular podcast; it's about as unobtrusive as advertising gets. Widespread circumvention of those ads could really hurt the ecosystem, which would be particularity frustrating given that podcasting appears to still be a viable space for small scale creators to do great things (e.g. San Charrington and the TWIML AI podcast)
TLDR cool demo but everyone should please think carefully about if or when to use this tool.
I think they address some of the reasoning behind this pretty clearly in the write-up as well?
> The potential risks of misuse raise concerns regarding responsible open-sourcing of code and demos. At this time we have decided not to release code or a public demo. In future work we will explore a framework for responsible externalization that balances the value of external auditing with the risks of unrestricted open-access.
I can see the argument here. It would be super fun to test this model's ability to generate arbitrary images, but "arbitrary" also contains space for a lot of distasteful stuff. Add in this point:
> While a subset of our training data was filtered to removed noise and undesirable content, such as pornographic imagery and toxic language, we also utilized LAION-400M dataset which is known to contain a wide range of inappropriate content including pornographic imagery, racist slurs, and harmful social stereotypes. Imagen relies on text encoders trained on uncurated web-scale data, and thus inherits the social biases and limitations of large language models. As such, there is a risk that Imagen has encoded harmful stereotypes and representations, which guides our decision to not release Imagen for public use without further safeguards in place.
That said, I hope they're serious about the "framework for responsible externalization" part, both because it would be really fun to play with this model and because it would be interesting to test it outside of their hand-picked examples.
That said, "unethical" was clearly a poor choice of words, as that's quite a loaded term. However, this technology is designed to systematically circumvent the mechanism by which many podcast creators earn money for their effort, and even if "unethical" is the wrong word, we should still discuss whether putting these technologies out, and using them, is a good move - particularly if we care about rewarding creators and supporting the podcasting industry. (And there's an argument that goes even further, around whether doing something like this is systematically violating an implicit contract with creators, even if it's more of a moral contract than a legal one).