Hacker News Reader

Comment by smcg | Hacker News Reader

What's stopping someone from paying to influence AI results?

How is that a different question than search results and which voter guides or topic editorials get highlighted first?

You have to disclose if a site is part of a PAC or candidate's committee due to FEC laws. This is not the case for LLM slop that often intermingles the two.

So yes, it's very very different.

Maybe these LLM companies should also disclose in every chat window on what political spending they have already done:

https://influence.citationneeded.news/

While this is true, there is a gaping hole in how foreign agents can amplify influencers on youtube and twitter. Probably even facebook, at this point.

An excellent reason not to use those sites, and certainly not for political decision making...

I guess FEC laws don't protect you from the pravda network. However, LLM might be not so good at ranking false information, if it is already massively embedded in their training data (LLM grooming).

Still it seems that this is currently not the case [0] (yet).

I guess the risk atm is rather some crazy billionaires embedding their believes in post training and system prompts. But honestly I rather fear tiktok in this regard.

[0] https://misinforeview.hks.harvard.edu/article/llms-grooming-...

Uncle Donald does. If you pay to influence against him, you are toast.

What’s stopping people from putting out biased voting guides that have nothing to do with AI?

Distribution.

Distribution is a solved problem. Social media, email/SMS campaigns, internet advertising, there are hundreds of ways to scale content distribution globally with zero AI.

While we can throw around hypotheticals and conspiracy theories about influencing AI results, we know for a fact that search engines and social media are heavily influenced, because there are giant industries around SEO and viral marketing -- plus the huge amount of ads and promoted content, which are literally paying to influence results.

Nothing. And it will happen big time once the AI companies become enshittified. Anthropic and OpenAI won't be able to produce the results that would justify a trillion dollar IPO without doing shady stuff.

The employees of the AI labs. They're a pretty vocal bunch, and would be very likely to quit, whistleblow, or both.

(xAI may be the exception here.)

Oh sure. The only whistleblower, Suchir Balaji. was murdered and the other employees just want to get rich (Brockman even is on record saying so).

A "vocal bunch"? They never depart from the party line.

Multiple high profile employees have quit... 99.99% have not. What has changed as a result?

We heard their concerns. My argument here is that any attempts within an AI lab to bias political answer would come to light.

Based on what? How could you possibly be sure?

Musk for example, has claimed in the open to push for fine-tuning to Grok that benefits him personally.

That's why I said "xAI may be the exception here."

Some have been vocal, a few have quit.

How has that changed how anyone interacts with AI?

1) People at large seem to prefer large stacks of cash to base morality

2) Employees will be tasked with writing a generic system that affects results, but only a handful of suits will know what it's actually going to be used for

Reply on news.ycombinator.com