"It really does disappoint me that so many brilliant colleagues—whose genuine breakthroughs I’ve profited from for years—would be so quick to condemn this newer, stupider way that I and others like me can make money off your life’s work, through stealing." Spot-on, as usual, from @mcsweeneys.net. - ThreadSky

adambecker.bsky.social • 22 days ago

"It really does disappoint me that so many brilliant colleagues—whose genuine breakthroughs I’ve profited from for years—would be so quick to condemn this newer, stupider way that I and others like me can make money off your life’s work, through stealing." Spot-on, as usual, from @mcsweeneys.net.

Comments

adambecker.bsky.social•22 days ago

LLMs and generative AI are fruit of the poisoned tree, and the plagiarism is one of the three main reasons why. The ecological impact is another (excellent work by @jamesmodonnell.bsky.social and @caseycrownhart.bsky.social for @technologyreview.com):

graham62.bsky.social•22 days ago

Very interesting read! Thanks!

adambecker.bsky.social•22 days ago

The final reason is the exploitation and psychological damage inflicted on the people who do content moderation work on these models, as described by @karenhao.bsky.social and @deepa.bsky.social here:

adambecker.bsky.social•22 days ago

So even putting aside the real problems with how people and companies are using LLMs, the bias inherent in the models themselves, & the hype that makes it nearly impossible for casual users to discern their true capabilities, LLMs still shouldn't be used for anything. They're ethically indefensible.

societyofonefifty.bsky.social•21 days ago

Apropos of nothing but a shared interest in this topic:

Do you have any sense of what the conversation about AI is like in China?

Do they have hysterical political figures promoting an AI arms race?

They seemed so cool with the DeepSeek release; like oh, here, were y'all interested in this?

societyofonefifty.bsky.social•21 days ago

Not sure they're ethically indefensible.

We think it's always possible to conceive a defense, however contrived, for giving people something they want. The shortest path to this being: not giving people something they want may cause them to suffer.

But we don't think most people want AI (yet).

john-gardi.bsky.social•22 days ago

It's the 'bros' picking the apple 𝘵𝘩𝘪𝘴 time! 😏

arniemi.bsky.social•21 days ago

They must acknowledge this plagiarism and compensate the affected parties. Probably a class action.

beesbuzz.biz•22 days ago

Also this article fails to capture that a lot of the added energy usage of AI isn't just from the AI models themselves, but the added costs shouldered by every website on the planet having to serve up a torrent of extra crawler traffic, most of which doesn't even respect "do not crawl" directives.

darksapiens.bsky.social•22 days ago

📌

Comments

Posting Rules

Reply