[ 3 / biz / cgl / ck / diy / fa / ic / jp / lit / sci / vr / vt ] [ index / top / reports ] [ become a patron ] [ status ]

/jp/ - Otaku Culture

Search:


View post   

>> No.32242318 [View]
File: 1.78 MB, 1546x2100, 1597027172578.jpg [View same] [iqdb] [saucenao] [google]
32242318

Always numba one

>> No.25671043 [View]
File: 1.78 MB, 1546x2100, 409966d628fa886bbe8c32cca9b383f4.jpg [View same] [iqdb] [saucenao] [google]
25671043

>>25670939
Very nice editing in this MV

>> No.25151768 [View]
File: 1.78 MB, 1546x2100, 409966d628fa886bbe8c32cca9b383f4.jpg [View same] [iqdb] [saucenao] [google]
25151768

>>25151360
I definitely don't think we need AGI (if making one is even feasible for humans) for a decent intelligent virtual companion/NPC experience. Right now we are good at encoding and decoding both voice and text (lots of problems that were troublesome in the past like dealing with the huge ways the same text can be structured is less of a problem now). What is truly missing is a reasoning system (something like the GPT-3 version of theorem provers and program induction). For example if you say "I hate peas" or "I really can't stand these green seeds", GPT-3 would encode it in a similar meaning space, and the new system could manage a fuzzy facts database and add a "user doesn't like vegetables" and perhaps compare with other facts like "vegetables are healthy for user" which will be used as the input for the GPT-3 decoding instead of solely the encoded text. It's hardly AGI, but it's something that is way beyond what we can do right now.

Navigation
View posts[+24][+48][+96]