Wouldn't it worked to get the AI to perform in a different environment and once it is shown it can learn and perform well, then the grounding takes place ?
Simon <div>-------- Original message --------</div><div>From: Alan Grimes <[email protected]> </div><div>Date:10-24-2015 22:04 (GMT-05:00) </div><div>To: AGI <[email protected]> </div><div>Subject: Re: [agi] HEY, NUMBSKULLS!!! DEFCON 2 SINGULARITY ALERT </div><div> </div> Duncan Murray wrote: > I am curious to see why people think that an AGI requires a high > resolution 3D graphical interface in which to operate / learn (at > least in the early stages) > > Not only is it a massive hit on computing/developer resources, but it > seems to be more of a distraction to AI projects. > > Couldn't an early stage AI work in a small 2D grid - the volume of > data is a lot more manageable. The issue is grounding. We need to train the AGI in a human-like environment so that we can apply standard psychological techniques on it and so that it can learn things that will be useful for us for it to know. -- IQ is a measure of how stupid you feel. Powers are not rights. ------------------------------------------- AGI Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/27714077-224a20ec Modify Your Subscription: https://www.listbox.com/member/?&; Powered by Listbox: http://www.listbox.com ------------------------------------------- AGI Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-f452e424 Modify Your Subscription: https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-58d57657 Powered by Listbox: http://www.listbox.com
