Comment by samwillis | Hacker News Reader

Comment by samwillis | Hacker News Reader

This is important to look back on in the context of what's happing now with AI tools. This story is obviously about the leak of the data publicly, but what it shows is the profiling that is available to corporations.

Search has exposed so much data about ourselves to the services we use with very little regulation on what they are permitted to do with it inside their own walls.

My fear with AI is that we are moving toward sending even more data to party services. Tools such a co-pilot (which I enjoy using) are a gold mine for behavioural analysis. The profiling that will be possible with these tools is extraordinary and we don't yet fully understand the implication.

It's because of this that I'm a massive proponent of "Local AI". We need to be pushing for the industry to adopt a local inference architecture asap. It needs to become the standard pattern as early as possible to reduce the risk of the AI revolution being a repeat of the invasive internet search and advertising industry.

Noise generation is the only answer the common person has against the giant corporate machine.

You need a personal "AI" that just does random searches unconnected with your life, constantly, in the background, and then injects this data into all the portals that are watching you.

Ultimately, their data will become dominated by noise, and ultimately useless to the point of severely destroying the value of the entire enterprise and data collection mechanisms in the first place.

No matter how many tools you make "local only" you're only a forgotten "send telemetry back to the mothership" checkbox away from being right where you started.

This might work at an individual level, but it isn’t scalable to the general population.

It’s hard to imagine any population-scale solution that doesn’t involve regulation.

The biggest problem with regulation (in my eyes) is that it thwarts competition between countries. E.g. if the US imposes restrictions on technology, innovation is incentivized to happen elsewhere. The EU has been bold on the privacy regulatory front with GDPR and the like, and has probably lost out on immeasurable monetary gains as a result. There’s a huge cost to regulation, but it works.

Why isn't it scalable? This could be a $19.95 device that you buy and just plug into your router. Instead of "Totally not the FBI VPN" why not "Noisy Gateway Service?"

I don't believe there's any worthwhile way to regulate this technology nor is there any particular reason to. The imbalance comes about because of spare electricity, which is truly the crux here, that we have so much to waste on dippy language models while also having so little to spare that some people have none... or possibly worse... electricity with random availability.

This you could actually regulate. I think the game is clear.

> Ultimately, their data will become dominated by noise

This was a plot point in a Neal Stephenson novel, "Fall [...]". It's been a while so I'm fuzzy on the details, but one of main characters floods the Internet with constant AI-generated content, but I can't remember if it was to drown out bad PR or break the anonymous identification mechanism ubiquitous in the book's world.

Until the noise you generate gets you on legitimate watch lists. One day it's going to search for a random product or compound that your neighbor two houses over just murdered their baby sister with and now you are an active suspect and get your electronics seized for a full search.

This argument boils down to "the government is too powerful and you shouldn't provoke it" while attempting to vouchsafe the legitimacy of metadata driven watch lists in the first place.

In my eyes this just strengthens the case.

The issue with AI is that people are going to be creating work product with it, and it doesn't require the extensive infrastructure that search has.

Google knows alot about your behavior - they can and have been able to correlate online behavior with health and meatspace actions to identify budding extremists or people at risk of addiction, etc. AI will bring that capability to business processes.

With the number of little companies that are springing up, it will become much easier for outside parties to figure out how instituions work. This capability exists, but it's gated by Google and Microsoft and they have drawn lines to protect the overall business. Some jackass will install a creepy AI tool to scrape outlook and salesguys will be able to get a profile of who makes what decision in a company, for example.

AI does take a ton of infrastructure. You need data collection and curation. Massive amounts of training hardware.

And a large infrastructure to ensure you can scale. No easy feat with the current GPU stack.

Are there not relatively tiny workloads that fall under the umbrella of AI? Or does the term itself inherently refer to intense workloads, like how the phrase "big data" refers to sets that can't be processed by typical means?

Training AIs is almost always done with massive data, as small data usually doesn’t have sufficient information content, statistically speaking, to build a good model from it. Certainly this is the case for the generative models.

I think AI refers to large models.

Smaller probabilistic models line linear regression, I’d call machine learning.

It takes a ton of infrastructure to maintain a sustained misinformation campaign.

Yet cloud providers happily sell resources and APIs to unethical companies. They rightfully don’t insert themselves into most legal business matters, with exceptions.

> Google knows alot about your behavior - they can and have been able to correlate online behavior with health and meatspace actions to identify budding extremists or people at risk of addiction, etc. AI will bring that capability to business processes.

not in the EU it won't.

if you can de-anonymize people from the data it's not anonymous, and collecting this data at all would be illegal in the EU without user consent, unless it's being used solely for the purpose of delivering the service.

I fully agree. If local AI takes traction, then we actually have a unique opportunity to take away some of that massive profiling, as some of what you use search engines for today can be done by AI. This may be why there's some fear-mongering around truly open-source AI and such by the big ones.

Reply on news.ycombinator.com