Hacker News Reader

Comment by sva_ | Hacker News Reader

So Google would need to build a discriminator that detects machine-generated content. It will be interesting to see these discriminators fight the generators of other big companies.

I'll be taking a front-seat row watching that show, if it happens. Perhaps in the future, we'll have to deal with discriminators that approximate some originality-index. Will be fun fighting with those algorithms, to interact with the internet as a normal user (to some extent we already do - proving that you're human is becoming more and more tedious.)

View on HN

In practice I think it's more Google now has another policy reason to banhammer prolific and irritating blogspammers than an arms race Google has a chance in.

Google isn't yet effective at detecting blogspam generated by naive scripts that simply swap words in the source material for other words in a thesaurus. I don't think they're going to start picking up continuity errors, factual errors or "weirdness" in GPT-3 - which often satisfies human readers - any time soon.

Google engineers can't even filter out GH or SO scraped sites like gitmemory, nor offer a way to let users block these sites. I'm not sure we should expect them to handle more advanced techniques like detecting word swaps any time soon.

s/can't/won't/

Google search has been poor for years. I think it's time to say can't.

That isn't how it works. Long term refusal to do it isn't evidence their engineers are incapable of it. Blacklisting sites is trivial. They've always been able to do it and they do it routinely to this day.

The reason those results remain in your searches is because sending people to those sites makes Google money, and by definition that is of greater value to Google than the quality and relevance of your search results as a non-paying user.

This filter list for uBO will likely be useful to you

https://github.com/quenhus/uBlock-Origin-dev-filter

The main reason I haven't tried this yet is I am concerned what would happen if every result is one of these spam sites that has fooled Google engineers for so long. Do I get a suspiciously blank page?

The other reason is I started using a search engine that cracked the code to let users rank and block sites (kagi) so I don't have to worry about it.

I am unsure about that but using GPT-3 in that manner would certainly trigger OpenAI's automatic or manual systems and would violate their ToS and your account would be locked.

With GPT-3, yes but how about other generators? There are lots of emerging services that spammers can use.

Does OpenAI actually have those systems? My impression is they are happy to take your money and just tell you not to do those un-nice things

Don't be evil (tm) ?

There is an approval process and when using the prototyping system it warns you about bad words.

Try searching for a speciality business in the middle of nowhere (where there are zero providers).

I recently did this, and ended up in a warren of sites like "10 best Widget repair in Smallsville, MO", where all the addresses were for nonexistent businesses in other states, with sort of plausible names tangentially related to widgets. (High plains frobulator sales, Mesa, AZ, or whatever).

I'm pretty sure a deep learning algorithm produced that corner of the internet. It definitely contained lots of referral links and phone numbers. Never managed to reach a human or figure out what the scam was.

These are usually lead gen sites. Done well they can take over smaller areas outside of major metro areas and drive sales leads to those larger companies that can travel vs the small local guy. I live in a small community and it has always been a problem usually started by the fact that the small local guy already has more work than he can handle and doesn't see the need to be online.

I've been noticing something similar with "phone-number look-up" sites.

I put that in quotes as I've entered numbers I know and find that what's being presented is apparently randomly-generated IDs.

The overt use is checking when unknown numbers call (or going through call history) and trying to get a sense of who the number is associated with.

The general location is usually plausible (so the generators are relying on local exchange numbers), but the specific street addresses and subscriber names appear generated, and specific street addresses may not actually exist.

You need to be computationally irreducible through self-attention and conscious self-interaction.

By the same criterion we may need to consider the Wattage cost of intelligence when granting rights to AI. Their kernels needs to be aligned with evolutionary wisdom encoded in our highest motivations.

a conversation for Q and Commander Data

Filmed in front of a live audience! [Applause]

Reply on news.ycombinator.com