Nothing can be trusted, everything can be improved
The powers of AI do not start and end with the singularity.
Cambridge Analytica proved narrative control for the mass was cheap
Now, control of the narrative for the individual is even cheaper
AI can be used against you to control your mind
Or you can use AI to free it
Which will you chose?
---
AI can be much more insidious than you realise, we can now replace all streams of information with content that is subtly altered in realtime, just for you. It is not hard to slowly integrate new ideas, manipulate narratives and create online subcultures when you have access to these streams.
Right now the only thing holding this back is our trust in TLS.
InsidiousAI demonstrates exactly how this might look and lets you experiment with different uses for this tool.
It's not all doom and gloom however, setting aside training set biases, AI can be used in real time to process the information you consume, provide insight in to areas you yourself have asked for help with, replace technical descriptions on the fly with ELI5s, explain the set of biases a text has and provide context on why a piece is highly emotive.
Use it to browse websites in foreign languages while still preserving the culture.
Use it to browse websites and understand your TikTok teen.
Use it to familiarise yourself with a world in which you no longer can trust anything.
Use it to censor the world around you and live in your perfect isolated thought bubble.
---
We were right to be paranoid before
We're more right to be paranoid now
Don't be part of someone elses plan, learn and grow
---
Usage:
You will need your own api key https://platform.openai.com/playground
To keep costs from ballooning and to preserve high demand inference cycles only Wikipedia and twitter have been integrated so far.
Please note that a few paragraphs all wikipedia pages and twitter will be sent to OpenAI for processing.
Everything else is local, I look forward to the day when I can remove the api call.