What does abating that trend look like? Most AI safety proposals I hear fall into the categories of a) we need to stop developing this technology or b) we need laws that entrench the richest and most powerful organizations in the world as the sole proprietors of this technology. Neighther of those actually sound better than people being paranoid weirdos about trusting text/video/voice. I think that's kinda where we need to be as a culture: these things are not trustworthy, they were only ever good as a rough heuristic, and now that ship has sailed. We have just finished a transition to treating the digital world as part of our "real" world, but it's time to step that back. Using the internet to interact with known trusted parties will still work fine, provided that some authentication can be shared out-of-band offline. Meeting people and discovering businesses and such? There will be more fakes and scams than real opportunities by orders of magnitude, and as technology progresses our filtering will only get worse. We need to roll back to "don't trust anything online, don't share your identity or payment information online" outside of, as mentioned, out-of-band verified parties. You can still message your friends and family, do online banking and commerce, but you can't initiate a relationship with a person or business online without some kind of trusted recommendation.
I don't think anyone has a good answer to that question, which is the problem in a nutshell. Job one is to start investing seriously in finding possible answers.
>We need to roll back to "don't trust anything online, don't share your identity or payment information online"
That's easy to say, but it's a trillion-dollar decision. Alphabet and Meta are both worthless in that scenario, because ~all of their revenue comes from connecting unfamiliar sellers with buyers. Amazon is at existential risk. The collapse of Alibaba would have a devastating impact on Chinese exporters, with massive consequent geopolitical risks. Rolling back to the internet of old means rolling back on many years worth of productivity and GDP growth.
> because ~all of their revenue comes from connecting unfamiliar sellers with buyers
Well that's exactly the sort of service that will be extremely valuable in a post-trust internet. They can develop authentication solutions that cut down on fraud at the cost of anonymity.
Even when it comes to people like our parents, there are things we would trust them to do, and things that we would not trust them to do. But what happens when you have zero trusted elements in a category?
At the end of the day, the digital world is the real world, not some seperate place 'outside the environment'. Trying to treat digital like it doesn't exist puts you in a dangerous place to be deceived. For example if you're looking for XYZ and you manage to leak this into the digital world, said digital world may manipulate your trusted friends via ads, articles, the social media posts they see on what they think about XYZ before you ask them.
Point a) is just point b) in disguise. You're just swapping companies for governments.
This tech is dangerous, and I'm currently of the opinion that its uses for malicious purposes are far better and more significant than LLM's replacing anyone's jobs. The bullshit asymmetry principle is very incredibly significant for covert ops and asymmetric warfare, and generating convincing misinformation has become basically free overnight.
