Oddbean new post about login | logout Then why wouldn't big tech do the following: 1️⃣ Overwhelm people with garbage content. 2️⃣ Offer paid plans to filter, not show them. 3️⃣ Still sell their data. I guess comfort could still keep the masses using these tools.
1. Big tech will build tools which will enable anyone to make garbage content. This is already happening. They will not be in control of what gets generated since there are many competing alternative “big tech” providers which offer similar products with somewhat varying techniques. 2. There is not a simple way to effectively filter this content. This will be especially challenging for big tech to filter since LLM content will be flooding in from different providers, different models, different “guardrails”, etc. Even filtering mostly-human/traditionally-automated content today can be challenging. LLMs make the problem much harder for big tech to solve. 3. It’s not clear to me what “selling data” means in this context. Are you imagining advertising systems like we have today or something else?
Under selling data I mean e.g. the information they get out of user prompts. They limit data generation to accounts, so that they can sell them as products. Then they can save your prompts under your account. I think people won't just use LLMs to generate data, but also to get to know certain answers. Also e.g.: if you are generating financial texts, most probably you are worth to advertise financial products to.
Certainly user prompts are a new type of data that could be included in a profile to target users with ads. But note, big tech companies don’t ever want to sell your data. They want to sell your attention, targeted to advertisers by using their proprietary systems on your data. They would prefer to keep the data for themselves and only charge advertisers for attention. The challengers are going to reshape the tools/interfaces to how data gets created/consumed/evaluated by users. There are more profound shifts afoot than just “more data captured by existing players”. Disruption at the interface layer is the biggest threat. Nostr plays a small role (for now) in this kind of disruption. It could play a larger role over time.
What do you mean under "interface layer"?
The web browser is an example of the “interface layer”. In the late 90s Microsoft tried to bundle Internet Explorer (their browser) into the Windows operating system, so that they could control the defaults (e.g. which “portal” or search engine most people use). Google fought this by building “search toolbars” that plugged into IE and other browsers and eventually launched the Chrome browser. It’s open source, but Google owns the “interface layer” (binary software distributed/installed on your computer which you launch). Another disruptive “interface layer” change happened years later when Apple launched iPhone/iOS and again controlled a new “interface layer” to computing. Apple uses this “interface layer” disruption to get $10s of billions of dollars from Google for the privilege to be the default search engine. A collection of open source nostr clients could become a new “interface layer” to the next wave of information and communication systems on the internet. As people seek higher quality information from more reputable sources with more control over their data, privacy, and attention there’s an opportunity to disrupt at the “interface layer” again. The largest strategic battlegrounds in business/tech occur at the “interface layer”. How do consumers choose what they do?
Good point @dk ! Data monopoly is better for them. They want your data, so they can monetize it without letting the control out! Thanks for crearing it! 🫡 Basically "infinite money" if they collect your data indefinitely, and they are the first point of contact for that data.