-Blackadder- posted:

First, we already have thinking machines that can outperform humans. Even a basic four function calculator can perform that specific function more efficiently and with less error than our "best and brightest" humans and certainly better than the average person.

But let's leave that point out and assume for the sake of argument that any AGI would come with human biases. So what? It would still make decisions that are vastly superior to the current reasoning process we employ in government and Kahneman explains why.

It's true that humans have a lot of things working against us when it comes to making rational unbiased decisions, but this is compounded a billion-fold into a global dumpster-fire of bad choices when we resort to snap judgement heuristics, which is pretty much always. The truth is given enough time, training, and a structured and rigorous methodology humans actually can work our way to pretty solid conclusions based on existing knowledge and evidence gained from design of experiments, but that rarely happens outside of something like Physics or other science research, and certainly not in modern politics. An AGI, inherent human biases included, would be able to make calculations using a database of all human knowledge to reach decisions in the most optimally human way possible, which is also far superior to what's in current use in government.

Like, I don't think people realize just how low the bar is right now in terms of Decision Science and Scientific Reasoning within our government, you could practically replace congress with plugged-in toasters and the outcomes would be better than they are now.

Or in other words, "It's displacement!"