Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

> Soon ChatGPT will start to weave ads into their output because they'll need to make $.

AI enthusiasts need to anticipate that. We're in the VC subsidy phase, but the hammer will drop sooner or later. If you think ads are bad on Google and Facebook now, just imagine a Google that has to spend 100x more on compute to service your requests.



Plus the incentives are totally fucked. It's bad enough with search, it's far worse with AI.

Nobody (referring to companies) wants the best model. Or the one that gives the right answer the first time 100% of the time. They want the model that's just good enough to keep you prompting, but just bad enough that you use a fuck load of tokens and see a million ads.

Unless they start making these things say more expensive, pretty soon developers are going to start seeing ads in the comments of their damn source code. Or worse, suggestions to use paid services to solve all your problems, because companies paid to have the LLM shill it's products.


> Or worse, suggestions to use paid services to solve all your problems, because companies paid to have the LLM shill it's products

and it's all going to be microsoft services shudder


Yes exactly. In a decade or two people will wistfully look back on 2025 era GenAI similar to how people currently remember when Google Search was great 10+ years ago. The AI enshittification will probably be even more dramatic though, in part because of the immense cost. Despite getting more abusive every year, Google ads are at least still identifiable if you know to look for them. Once AI training weights start getting heavily manipulated for corporate and political reasons things could get pretty ugly.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: