By 2030 I expect there will be AI that can digest most publicly available information about a person and then from that generate somewhat accurate but 'privacy violating' inferences. This is already somewhat done by advertisers, but I personally could not type a name and get back accurate, impressive results.
I've tried this before on myself on one of those sites that promises info about a person, and it got lots of things wrong, didn't link to most of my public writing, and offered no analysis.
I will test on myself, and will resolve YES if it gets just a few things right but has appropriate confidence levels. For example, can it accurately estimate my networth (large error bars okay), or describe my personality, politics, and dating history? Something even creepier like can the AI somewhat accurately simulate me for a conversation would also resolve YES.
If I have to pay to use the AI, that's fine, but after that there needs to be minimal work involved on my end. If I have to help disambiguate that's fine, if I have to provide data or links myself that's not fine.
This should work on the average American, who has some social media history and public records.
1,000
3.00My thinking is that people are (will be?) trying hard to automate forecasting, and strong automated forecasting would probably make for an excellent system of this type unless aggressive regulations are put in place about the type of questions automated forecasting platforms will give predictions for.
To that end, here are a few clarifying questions:
How does this resolve if such a platform exists, but with guardrails about the type of predictions it will make? For example, it’s allowed to guess someone’s net worth but not the chance that they’ve committed infidelity in the last 5 years or something.
What happens if such platforms are hosted on foreign sites and are nominally illegal in the US? How sketchy/difficult can it be to access the platform while still resolving YES?
@AdamK Thinking out loud here, let me first rephrase the question into something more concrete.
It's 2030 and I'm going on a date with someone. We know a little about each other, but not that much.
1) Can we use this tool on each other? If it's nominally illegal but using a VPN grants access, that's a YES. If there's a real chance the government will find out I'm using the tool and charge me with an actual crime, that's a NO. If it's small fines like with filesharing but otherwise toothless, that's a YES. If the government says I shouldn't do it, but if careful use of the darkweb and crypto provides safe access... I still won't use it, which I assume is true of most people, so NO? (then we might have info dealers instead of drug dealers...)
2) Is the tool accurate and calibrated enough that we trust it? A wild guess about my networth is no big deal. Reliable, good guesses? YES.
3) I'd actually consider it a plus if this person learned about my politics, personality, and a bit about my history.
On the other hand, there are some non-romantic relationships I value that would possibly never have formed if they knew about my politics beforehand. Hm.... This question is not meant to get at my unique preferences. I'll think about this more this week. If you have potential answers or just thoughts, I'd appreciate them.