Hi All– first time poster. I have worked in SaaS for nearly 15years. I became “radicalized” on consumer data privacy from my first job experience, where I worked closely with large enterprises like Coca-Cola, Nike, even Phillip Morris(yikes) to help them collect and manage first-party data so they could sell their products and peddle influence(for lack of better phrasing). During this time I observed a few things:
Virtually all businesses treat privacy as an obstacle rather than an ideal. It’s something to be worked around rather than respected.
Data increasingly fuels the business. People increasingly are the product
Again, these observations were ~15yr ago, so not mind-blowing statements now, but it started me down the path and shaped where I am today. One of the ideas that was new/scary at the time was harvesting social profiles for consumer targeting. It was really the first time that reliable information about a person was so centralized and accessible. The social networks were happy to hand these consumer profiles out to anyone who would use them to buy ad spend on their sites. To me and many others, this is where social networks began to spiral from something fun and community oriented to algorithm-driven targeting machines. In many ways our personal profiles were used to alienate us from the communities we originally joined social media to enjoy and pushed us into separate corners of the internet surrounded by bots and advertisers looking to profit on us. This type of data-harvesting and targeting played a big role in what the internet looks like today– virtual tribes and echo chambers that we’ve been sorted into by various algorithms.
Fast forward to today- with the advent of AI, I see many of the same signs that I saw 15years ago. This time, however, I believe the risk to be more dangerous, because it has the power to expose our private lives in ways traditional data collection has, up until this point, been unable to crack. AI has become an easy companion. It makes us productive, and in exchange we tell it more about us. It learns from the things we say, the way we say it, and even infers things about us from the things we don’t say. As agentic workflows come to life, we’ll give our AI companions access to our emails, calendars, smart home devices, health devices because, why not? They’ll make our lives easier. At some point we may even gift access to our financial accounts and transfer purchasing authority over to these companions.
AI will centralize our lives under its umbrella in exchange for productivity. What most people don’t consider is that it can and will be used by others to target and ensnare us more deeply in various webs of influence(whatever they may be). If you look at Anthropic, OpenAI, Google, Amazon, etc. all of them are making moves to own the AI memory layer for you and your family. This is the single most disturbing thing to me about the current state of affairs. I think that the dystopian, almost sci-fi question of “will we ultimately own AI, or will AI own us?” is very real and should be taken seriously. I believe the future of the free internet depends on us building distributed AI with self-sovereign ownership over the memory layer, not centralized AI owned by few. Curious to hear some reactions to my thoughts.
It’s really not AI I am scared of and mistrust, it’s the perverse incentives behind it, human stupidity, and greed.
We’re not evolved enough to wield such a power when it truly becomes something to contain much like radiation. Or at-least almost everyone in power are not evolved enough. Rationality is not rewarded. Positive projections, however false, explained with charisma is.
Agreed. I think the last part of my post came off as people vs. autonomous AI super intelligence. I’m most concerned about a few people controlling the AI that is used by many.
While it may not necessarily be the cause of the centralization of the net, the rise of so-called “AI” and general powerlessness to do much about it is absolutely a symptom of this problem.
As for this, there’s absolutely evidence to the contrary, in terms of local LLM usage, and processor NPU capabilities. Certain queries do not require large models and most certainly can be done decentralized on the user’s device. This saves a round trip, and doesn’t disclose any data.
Absolutely. I think Locally hosted LLM with Cloud PII redaction for any inference that has to hit a frontier model is not only prudent but necessary for those that want to maintain privacy going forward. The tradeoff of privacy for productivity doesn’t have to happen, but most people doin’t know that.
NPU is literally in every new processor. As time advances they will get more powerful and be able to handle a lot of the simple tasks people throw at them.
There’s always going to be room for both. Some things a small model can do quite easily, like semantic searching eg “find me the invoice where I bought a shoe” should never leave the system. You might not have the word “shoe” in the file, but maybe “sneaker” instead. A small language model is perfect for this. You can use a sliding window that clears as it goes since it doesn’t need to reference those documents again. This can all be done easily on a small NPU.
There are plenty of other examples where you need huge context windows, especially when dealing with massive datasets. In those cases, it makes more sense to use the cloud, especially for data that isn’t sensitive.
Contrary to popular belief, public LLMs such as ChatGPT and Gemini do not train on active conversations in real-time. Training on this type of data would lead to model collapse. For example, a model trained on low-quality data from someone who thinks that an LLM is intelligent and now their boyfriend would not be useful to anyone.
With advancements in model compression it’s not out of the question to imagine a not-so-distant future where you can run a trillion parameter model locally on the equivalent of a raspberry pi. That’s what has me most excited.
I really do believe the long-term future of AI is local. History tends to move in that direction, technologies that once belonged only to governments, universities, or massive corporations eventually become personal. Computers used to fill up whole rooms & now they fit in our pockets and have incredible capabilities. AI feels like it’s still in its “mainframe era” and it’s gonna get get messier before it gets better, but I think the endpoint is clear and it isn’t rented access to a server farm far away.
Personally I’m hoping that AI is eventually relegated back to only being used in embedded and specific systems that actually benefit humanity such as protein folding, instead of generating endless slop so that Sam Altman, Dario Amodei and Jensen Huang can add another zero to their net worths. I think the only winning move with AI is not to play, or in other words, to reject it in your life where you can.