4 ms·
Well yeah. You could also be afraid of screwdrivers. After all, someone could stab you with one. So on the one hand you could fear the screwdriver and try to
by Falkon1313 4y ago
Well yeah. You could also be afraid of screwdrivers. After all, someone could stab you with one. So on the one hand you could fear the screwdriver and try to eliminate them. But really, it's the stabby person you should be afraid of, not the tool. If not one tool, they'd just use another.
- onethought 4y agoWithout derailing this discussion… Gunlaws have proven to be very effective. It is a compelling counter example of: no people don’t just use another tool if you block/ban one.
- xyzelement 4y agoI haven't encountered this concept before - where has this been proven very effective?
- somewhereoutth 4y agoPretty much every developed country outside the US. Along with universal healthcare, it is considered a marker of a well functioning state (monopoly on violence).
- xyzelement 4y agoIs that's the same as pointing to how the laws have proven effective? I would expect something like an apples to apples before and after comparison. Also, as someone who grew up in an oppressive state, government monopoly on violence is far from a blessing.
- onethought 4y agoAustralia 1997 is when they introduced strict gun laws. Go check the stats on gun violence. Also go see that other violent crime did not increase either. Hell, ask chat gpt to explain.
- pmoriarty 4y ago3d-printed guns may make gun laws obsolete before too long. Besides, the laws have only stopped some of the rabble. They have not stopped governments from developing ever more effective and deadly weapons. Technological progress in weapons development has not stopped, and it's that kind of progress (rather than mere gun ownership) that's more akin to AI development.
- onethought 4y agoSo like human cloning? Or human experimentation? Or biological weapons… banning them has no effect?
- simonh 4y agoScrewdrivers don't suffer from the alignment problem. The issue with LLMs is that the criteria for success in their training are proxies for the actual behaviours we want to achieve. Therefore there is the risk of misalignment between what the LLM is actually doing and what we intended it to do. This is not a theoretical risk, petty much all the misalignment pathologies AI researchers have been worrying about are exhibited by ChatGPT and other such models, in spades. Basically very often the models find it easier to deceive us into accepting a pathological response rather than generating a genuinely correct response. Robert Miles' Youtube channel is required viewing for anyone with an interest in this.