On Thu, Sep 18, 2008 at 3:26 PM, Linas Vepstas <[EMAIL PROTECTED]>wrote:
> > > > I agree that the topic is worth careful consideration. Sacrificing the > > 'free as in freedom' aspect of AGPL-licensed OpenCog for reasons of > > AGI safety and/or the prevention of abuse may indeed be necessary one > > day. > > Err, ... but not legal. What do you mean? The SIAI and Novamente hold the copyright for OpenCog code, and are perfectly within their legal rights to change the terms of the license of SIAI-distributed source code. Of course changes cannot be retroactively applied to source code already distributed, and there are no plans to make any license changes, but such changes can be made perfectly legally. Also of course the SIAI would need to be in a position of significant influence (like, say, employing key developers and driving key progress or holding contracts with corporate/government users or exerting influence over commercial policy or government regulation, etc.) for any license changes to be relevant in a software economy where anyone with sufficient skills and influence could maintain a fork using the old license terms. > > > > One of many obstacles in the current legal framework worth considering > > is that machine-generated things (like the utterances or self-recorded > > thoughts of an AGI) are uncopyrightable and banished into a legal no- > > mans-land. There is simply no existing legal framework to handle the > > persons or products originating from AGIs. > > Law is built on precedent, and the precedent is that works > produced by software are copyrightable. If I write a book > using an open-source word-processor, I can claim copyright > to that book. If I press a button that causes an open-source AGI to write > a book, (possibly based on a large collection of input data > that I gave it) then I can claim ownership of the resulting work. > Original works produced by software as a tool where a human operator is involved at some stage is a different case from original works produced by software exclusively and entirely under its own direction. The latter has no precedent. > No, the crux of the problem is not that the output of an AGI > isn't copyrightable ... it is, based on the above precedent. > The crux of the problem is that the AGI cannot be legally > recognized as an individual, with rights. But even then, > there *is* a legal work-around! Claiming a copyright and successfully defending that claim are different things. I agree that the non-person status of [some future] AGI is a bigger problem. Of course, a trans-human AGI is .. err.. will defacto find > that it is not bound by human laws, and will find clever > ways to protect itself, I doubt it will require the protection > of humans. Recall -- laws are there to protect the weak > from the strong. The strong don't really need protecting. > AGIs will likely need protection from other AGIs, and I expect they will create AGI-society legal frameworks, perhaps similar to or originally based on human laws. > > I'm not worried about people enslaving AGI's; I'm worried > about people being innocent bystanders, victimized > by some sort of AGI shootout between the Chinese > and American CIA -built AGI's (probably by means of > some propaganda shootout, rather than a literal guns > and bombs shootout. Modern warfare is also > homesteading the noosphere) I believe that James's concerns cover both AGI mental torture (coercing or tricking a conscious entity into behavior which is sociopathic or criminal or otherwise immoral) as a heinous act in itself and also the 'crossfire' concerns you raised. -dave ------------------------------------------- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244&id_secret=114414975-3c8e69 Powered by Listbox: http://www.listbox.com