Hacker News Reader

Comment by TillE | Hacker News Reader

Google is vague about this, but the easiest solution to protect the quality of the ngram viewer is simply to only use books from known, non-vanity publishers.

View on HN

Everyone jumping into AI, but there's going to be a business opportunity for curated sources of information with proof that actual human beings created it.

We need to stop looking at every negative externality caused by GenAI as a "business opportunity".

Sweeping the streets of confetti and trash after New Year's Eve is not a lucrative job.

The negative externalities are here regardless. We just spent the last two decades letting big tech get away with whatever they want, and they've gobbled up the internet and now hold all of the cards. The ocean of AI sludge getting shat all over us is only going to accelerate and for the majority of people will be the norm, there'll be no competition on cost or ease of access.

Our best hope is to create islands of by-humans for-humans media, software, communities, etc. They'll be "premium" in the sense that they'll more expensive and more exclusive, but for the people who are opposed to the fuzzy auto-completing away of our humanity it'll be worth it.

I do think that subcultures will develop with human-only values that aren't built around businesses, and the two will have a lot of interplay.

It's more akin to filtering or stopping email spam, which can be somewhat lucrative.

The business opportunity created for prosthetic manufacturers is not a sufficient justification for the mass manufacturing and deployment of landmines.

How can this be proved? Any human author can copy/paste snippets out of ChatGPT.

Suppose this is the way we want to go (many people probably won't agree for good reasons), Google is not going to put in the effort to identify reputable publishers.

Honestly I am a bit surprised that Google Books is still around -- does it bring a lot of revenue? Why hasn't Google got bored with it by now?

My guess: it's used as a high quality source of training data (well, mostly anyways) since papers like Textbooks Is All You Need showed that that can be a treasure troth. Offering this as a "public service" may just be a smokescreen for regulators.

Oh, so the gatekeepers serve a purpose.

Curation serves a purpose, Google is one, there are more

We want competition and options in this space. It's one of the things I really like about Bluesky, which I only discovered recently. They enable competition and options for all core components of social media

Gatekeeping as a term has lost its meaning. These days it’s used to refer to any kind of quality control.

Reply on news.ycombinator.com