Hacker News Reader

Comment by b00ty4breakfast | Hacker News Reader

What I am worried about (and it's something about regular internet search that has worried me for the past ~10 years or so) is that, after they've trained a generation of folks to rely on this tech, they're going to start inserting things into the training data (or whatever the method would be) to bias it towards favoring certain agendas wrt the information it presents to the users in response to their queries.

View on HN

> after they've trained a generation of folks to rely on this tech ... bias it towards favoring certain agendas

previously, this happened with print media. Then it happened with the airwaves. It only makes logical sense that the trend continues with LLMs.

Basically, the fundamental issue is that the source of information is under someone else's control, and that someone will always have an agenda.

But with LLMs, it's crucial to try change the trend. IMHO, it should be possible for a regular person to own their computing - this should include the LLM capability/hardware, as well as the model(s). Without such capabilities, the exact same will happen as has in the past with new technologies.

> it should be possible for a regular person to own their computing

And regular persons will not care about this and will select a model with biases of anyone who they deem "works better for me at this one task that I needed".

Just like you said:

> previously, this happened with print media. Then it happened with the airwaves. It only makes logical sense that the trend continues with LLMs.

I wish it wasn't so, but I have no idea how to make people care about not being under someone's control.

I worried about this a lot more at the tail end of 2003, when OpenAI's GPT-4 (since March) was still very clearly ahead of every other model. It briefly looked like control of the most useful model would stay with a single organization, giving them outsized influence in how LLMs shape human society.

I don't worry about that any more because there's so much competition: dozens of organizations now produce usable LLMs and the "best" is no longer static. We have frontier models from the USA, France (Mistral) and China now.

The risk of a model monopoly centralizing cultural power feels a lot lower now then it did a couple of years ago.

Model competition does nothing to address monopoly consolidation of compute. If you have control over compute, you can exert control over the masses. It doesn't matter how good my open source model is if I can't acquire the resources to run it. And I have no doubt that the big players will happily buy legislation to both entrench their compute monopoly/cartel and control what can be done using their compute (e.g. making it a criminal offence to build a competitor).

Model competition means that users have multiple options to chose from, so if it turns out one of the models has biases baked in they can switch to another.

Which incentivizes the model vendors not to mess with the models in ways that might lose them customers.

I don't think anyone considers biases more important than, say, convenience. The model that only suggests Coca–Cola brands will win over the one that's ten times slower because it runs on your computer.

I don't think model competition is necessarily the fix to this issue. We're not even sure if the setup as it exists today will be the norm. It could be that other entities license out the models for their own projects which then become the primary contact point for users and LLMs. They are obviously going to want to fine-tune the models to their use-case and this could result in intentional commercial or ideological biases.

And commercial biases wouldn't necessarily be affected by competition in the way that you're describing. For example, if it becomes profitable for one of these companies to offer to insert links to buy ingredients at WalMart (or wherever) for the goulash recipe you asked for that's going to become the thing that companies go after.

And all of this assumes that these biases will be obvious rather than subtle.

Absolutely. Like most things on the Internet, it will get enshittified. I think it is very likely that at some point there will be "ads" in the form of the chat bot giving recommendations that favor certain products and services.

> to bias it towards favoring certain agendas wrt the information it presents to the users in response to their queries.

Do you mean like Grok is already doing in such a ham-fisted way?

This is already happening. People are conditioned to embrace capitalism, where a small percentage of the population are born into the owning class, and a majority who labour.

Being told how my grandma had problems and was eventually told to shut down her knitting production (done in free time in addition to regular work) by police in the Communist Poland, I believe that it's better to have somehow upgraded capitalism then try to build a good communism just one more time.

It still got her enough extra buck to build a house in the city after moving out from the village.

Communism is neither the opposite of laissez-faire capitalism nor the only alternative.

The opposition to capitalism have such a disastrous track record, economically and in terms of body count, that embracing capitalism is far more sensible.

I'm not saying that the other systems, by which I assume you mean the various marxist political projects, are good (and we won't even get into how many of those alternatives were actually not-capitalism) but I think to dismiss the "body count" of capitalism while simultaneously ascribing all deaths under those alternative systems as the direct result of {otherSystem} is extremely disingenuous. Doubly so given that modern first-world capitalism often outsources the human cost of it's milieu to the third world so that middle-class suburbanites don't have to see real price of their mass-produced lifestyles.

Modern Western countries mostly drifted towards a mix of capitalism and social democracy.

"modern first-world capitalism often outsources the human cost of it's milieu to the third world"

This is a bit of "damned if you do, damned if you don't".

If you don't do any business with poorer countries, you can be called a heartless isolationist who does not want to share any wealth and only hoards his money himself.

If you do business with poorer countries, but let them determine their own internal standards, you will be accused of outsourcing unpleasant features of capitalism out of your sight.

If you do business with poorer countries and simultaneously demand that they respect your standards in ecology, human rights etc., you will be accused of ideological imperialism and making impossible demands that a poorer country cannot realistically meet.

Pick your poison.

The alternative systems were just as willing to plunder their satellite states and the third world IIRC as the capitalists were so it would be an equal demerit for both, I'd think?

I think that's called feudalism. Maybe our reality doesn't work like it's named and we are starting to have other system despite what we are calling it.

Reply on news.ycombinator.com