Large language models (LLMs) like GPT-4 can identify a person’s age, location, gender and income with up to 85 per cent accuracy simply by analysing their posts on social media.
But the AIs also picked up on subtler cues, like location-specific slang, and could estimate a salary range from a user’s profession and location.
Reference:
arXiv DOI: 10.48550/arXiv.2310.07298
Explaining what happens in a neural net is trivial. All they do is approximate (generally) nonlinear functions with a long series of multiplications and some rectification operations.
That isn’t the hard part, you can track all of the math at each step.
The hard part is stating a simple explanation for the semantic meaning of each operation.
When a human solves a problem, we like to think that it occurs in discrete steps with simple goals: “First I will draw a diagram and put in the known information, then I will write the governing equations, then simplify them for the physics of the problem”, and so on.
Neural nets don’t appear to solve problems that way, each atomic operation does not have that semantic meaning. That is the root of all the reporting about how they are such ‘black boxes’ and researchers ‘don’t understand’ how they work.
Yeah but most people don’t know this and have never looked. It seems way more complex to the layman than it is because instinctually we assume that anything that accomplishes great feats must be incredibly intricate