Slight unintended consequence is it creates a significant incentive for politicians to allow their expenses to be picked up by someone else. There are rules in many democracies where "gifts," over small amounts (~$50) must be reported to ethics commissions and scrutinized for opportunity by the opposition, media, and by extension, the public.
In a simple risk game of likelihood vs. consequences, expensing a meal in an upper end restaurant will %90 draw political consequences from an AI, with a high degree of uncertainty in the cost to fight it, vs. accepting the hosts generosity, with a sub-%5 chance of consequences occurring like having to pay your share, or a minor sanction some years into the future.
I'd argue most politicians necessarily make risk and cost/benefit calculations before ethical signalling ones, and the indiscriminate application of rules in this case creates inverted incentives that demoralize the situation, and will cause them to start accepting insignificant bribes.
The economic value of "AI" is mostly to obfuscate and symbolically defray the chain of individual human accountability. While this may be working for the "good guys," in the public interest, it's really about protecting the people who are using what is essentially a Bayesian black box as an ethical proxy for their personal interest. It's fascinating and cool, but it's also a bit disingenuous to accept magical pretexts and other ethical proxies because we perceive them to be "on our side."
I don't get most of the argument here, what are you trying to say?
Politicians are already incentivised to get their bribes indirectly, that's a natural consequence of doing something illegal while expecting to have your actions recorded. But beyond that I don't understand, are you saying that because the AI allows for such efficient oversight that politicians will be unable to hide large bribes and be forced to accept "insignificant bribes"? But even if the only effect is that corruption is much cheaper, what is the downside to that? At least more people can afford it then.
> it's really about protecting the people who are using what is essentially a Bayesian black box as an ethical proxy for their personal interest
How is fighting corruption best described by "personal interest"? The only way I can make sense of this is of you hold the belief that rule of law is not fundamental for a just society, and the circumvention of the law that corruption represents somehow makes societies more just.
What I have said is that this particular AI approach changes the risks and incentives for politicians so that they are more likely to commit a greater number of small, but more real offenses, than the few outward anomalies the AI flags.
Which is worse? Expensing a bottle of wine, or having a vendor pick up the tab for an entire meeting? I would argue the latter has more impact on perceptions of corruption, yet this demoralized AI approach disproportionately punishes the former.
There is a style of argument that I'm starting to think of as "argument from tyranny," which begins with a denial of the coherence and legitimacy of the speaker, and ends with a defamation of character. They call it "sealioning," on the internets, but it bears more scrutiny, because it is part of why people are reduced to using pseudonyms.
And the alternative is what? Just let them spend whatever they want? How can politicians in Sweden do the same job without spending exorbitant amounts of money? They clearly don't need this much money.
In a simple risk game of likelihood vs. consequences, expensing a meal in an upper end restaurant will %90 draw political consequences from an AI, with a high degree of uncertainty in the cost to fight it, vs. accepting the hosts generosity, with a sub-%5 chance of consequences occurring like having to pay your share, or a minor sanction some years into the future.
I'd argue most politicians necessarily make risk and cost/benefit calculations before ethical signalling ones, and the indiscriminate application of rules in this case creates inverted incentives that demoralize the situation, and will cause them to start accepting insignificant bribes.
The economic value of "AI" is mostly to obfuscate and symbolically defray the chain of individual human accountability. While this may be working for the "good guys," in the public interest, it's really about protecting the people who are using what is essentially a Bayesian black box as an ethical proxy for their personal interest. It's fascinating and cool, but it's also a bit disingenuous to accept magical pretexts and other ethical proxies because we perceive them to be "on our side."