Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

The worry is not human-level unaligned AI, but superhuman-level one.

There is no superhuman-level corporation yet.



I'd argue that organizations (whether corporation, government, or otherwise) are by definition superhuman. What economic incentive would there be to form companies if an individual could do it all better? Clearly, people working in concert are more effective than those working alone.


By this definition 100 humans pushing a stone have superhuman abilities. This is not what superhuman means in an AI alignment context.


> By this definition 100 humans pushing a stone have superhuman abilities.

Yes, that is accurate. A person with the strength of 100 people would be superhuman. A group of more-than-one humans has more-than-one-human abilities.

> This is not what superhuman means in an AI alignment context.

It is though, at least partially. Chess AI are great at chess because they play a shitload of it. Part of the power of AI is just being able to do the same things we do at vastly increased scale. Even if it could only do that, it would be dangerous.


there is another question moralists are avoiding. Is it moral to keep intelligent creature as a slave? How about forever? So far many realize it may be not safe.


There's a huge leap between intelligent and sentient, and approximately 100% of human labor could be satisfied somewhere in between.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: