<html><head></head><body><div><span data-mailaddress="alex.urbanec@gmail.com" data-contactname="Asdd Marget" class="clickable"><span title="alex.urbanec@gmail.com">Asdd Marget</span><span class="detail"> <alex.urbanec@gmail.com></span></span> , 27/10/2014 7:21 PM:<br><blockquote class="mori" style="margin:0 0 0 .8ex;border-left:2px blue solid;padding-left:1ex;"><div>AI could be incredibly dangerous if we don't get it right, I don't think anyone could argue against that but I never see these types of articles discuss the methods and models we are attempting to develop for "Friendly AI."</div></blockquote></div><div><br></div><div>http://slatestarcodex.com/2014/08/26/if-the-media-reported-on-other-dangers-like-it-does-ai-risk/</div><div><br></div><div>(Actually, there are surprisingly many professional AI people who argue against the danger of successful AI.)</div><br><div><blockquote class="mori" style="margin:0 0 0 .8ex;border-left:2px blue solid;padding-left:1ex;"><div> In my opinion, we should be working harder on concepts like Yudkowsky's Coherent Extrapolated Volition (<a href="https://intelligence.org/files/CEV.pdf" title="https://intelligence.org/files/CEV.pdf" target="_blank">https://intelligence.org/files/CEV.pdf</a>) to ensure we aren't simply ending our species so early in our life cycle.</div></blockquote></div><div><br></div><div>Well, Eliezer would be the first to admit CEV is flawed and by now obsolete. Nick's book has a chapter on value loading, and the MIRI crowd is getting into ever more esoteric branches of logic to make something that is CEV-like. </div><br><br>Anders Sandberg, Future of Humanity Institute Philosophy Faculty of Oxford University</body></html>