scott-smith_us
Joined 8 years ago
Joined 8 years ago
In the foreign invasion (or US govt goes full fascist) case, people can remove these devices.
Meanwhile many if not most high-speed chases won't happen any more, because the person being chased didn't plan ahead.
There will still be getaway cars, etc. where the device was removed. That's no different to what we have now, and it'll be an extra violation to charge them with when they're caught.
Since driving is, and has always been a privilege and not a right, I don't see a problem with having a remote kill switch that cops can use in specific cases.
*Freedom of movement* doesn't require that you are able to drive your own car. It's not *new* either. Get pulled over for a DUI, and you won't be allowed to drive for a while.
It absolutely does not mean that they can't tell you you're not allowed to drive. Driving is, and always has been a privilege in the US; not a right.
Some exceptions:
- Commercial use of (recognizable) photos of people in public is regulated.
- Following someone around or getting in their face with a camera is harassment.
Of course, your right to take photos or videos stops the minute you step onto private property; the owners are within their rights to disallow it.
Also, the tools seem to be of very low quality. For example, many of the sliders are set to initial values that are 'tuned' to give a typical example. Most sliders have fairly large ranges for which (it seems) only a small sub-range typically yields interesting results. Their 'Random' button doesn't seem to be weighted toward these subranges, so the result is usually a bunch of extreme settings that all look similar.
- You never know how it's going to be judged (style, efficiency, error handling, abstraction, thread-safety, type-safety, commenting, etc).
- You're given a very short time-limit, but well-designed, maintainable software is almost never written like that.
- etc., etc.
From the moment I first savored the delectable aroma of your signature charbroiled burgers, I knew that my destiny was intertwined with your esteemed establishment. The synergistic blend of your brand's values and my personal ethos is a match made in culinary heaven, a union that promises to revolutionize the gastronomic landscape.
The prospect of contributing to Hardee's renowned legacy of taste innovation and unparalleled customer satisfaction resonates deeply with my quest for self-actualization in the fast-food industry.
Each meal served, every customer smile elicited—these are the very keystones that embolden my commitment to fulfilling my dream of championing a new era of succulent satisfaction. With utmost fervor, I implore you to consider my application, for it is not just a role I seek, but a calling I am destined to embrace with zest and zeal."
I went ahead and filled it out while I was there, but I'd already decided not to move forward by the time the interview was over.
But most of the arguments aren't about bad actors. The basic mechanisms of AI give rise to a bunch of behaviors that are problematic with weak AIs and possibly extinction-level with super AIs. This isn't behavior that we just suppose they might have, given common-sense understanding of mammal psychology. This is behavior we've observed in many, many systems, that is a consequence of how AI works.
For example:
- AI seeks what you reward it for (i.e., what it "wants"), which is often exactly what you said vs. what you really meant.
- AI is incentivized to prevent you from turning it off, or from updating (fine-tuning, correcting) its reward function.
- A sufficiently smart AI will deceive you into believing it's doing what you want if it knows that you'd stop it before it gets what it wants.
- A sufficiently smart AI could modify its own code, or create sub-agents to get around safety measures that prevent it from reaching certain "better" solutions.
- Etc, etc. These are really interesting videos explaining the different issues: https://www.youtube.com/c/robertmilesai
The list above is based upon provable behavior given the current mechanisms in AI. The only thing that makes this a future problem is that we don't currently have AI systems that are as smart or smarter than us. The history of AI research is notable in that estimates about the pace of progress have been all over the map. No one really knows how many more discoveries or refinements are required before AGI becomes possible.
Once they get many times smarter/quicker than us, we need to be sure AIs are built NOT to want to do these things, because we're unlikely to be able to outsmart them.
No one is claiming that alignment is a well-defined thing with crisp edges. They're saying that the mechanisms of AIs will favor solutions without regard to any specific constraints we haven't explicitly stated.