A nyelvmegértés már nehezebb terep, mint az utánzás. Utánoznak a reneszánszukat élő chatbotok is, a Stanford együttműködő ágensei, valamint az MIT Számítástudomány és Mesterséges Intelligencia Laborjának (CSAIL) egyik fejlesztése viszont a nyelvmegértésre fókuszál: felvett beszédet képekhez kapcsoló rendszerrel jutnának el a teljesen automatizált beszédfelismeréshez. A két különböző ideghálót használó rendszer tematikusan kapcsolódó képek és az e képeket leíró szövegek (amelyeket egy nagy audiofelvétel-gyűjtemény tartalmaz), viszonyát elemzi. Az akusztikai jegyek képekhez társításából – felügyelet nélkül – tanulja meg, hogy milyen hangzó elem milyen gépelt szónak felelhet meg. Lényegében itt is az a cél, hogy úgy tanulja meg a nyelvet, ahogy az ember, de ne csak tanulja, hanem értse is.
Innen próbálnak még kísérletibb, ismeretlenebb terepek felé mozdulni többen, köztük az egyébként nem nyelvész, hanem robotikus Mordatch is.
Nyelvet teremtenek az ágensek
A Mordatch általa épített egyszerű kétdimenziós virtuális világban zöld, piros és kék körökkel megjelenített ágensek kényszerből hozzák létre saját nyelvüket. Feladatokat kell végrehajtaniuk, együtt kell működniük hozzá, közben jutalmat és büntetést kapnak. Segítségnyújtáskor tanulják meg a nyelvet, hogy elmondhassák egymásnak, hogyan jutnak el gyorsabban valahova. Véletlenszerű karaktereket társítanak a miniatűr világban való tájékozódáskor megtanult egyszerű fogalmakhoz, egymáshoz, helyekhez, tárgyakhoz és cselekvésekhez.
![](data:image/jpeg;base64,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)
A fentebbi képen a piros ágens a piros ponthoz kapcsolódó szót mond (t=0), majd, hogy „menj valahova” (t=1), ezt követően pedig, hogy „zöld ágens” (t=2). A zöld ágens meghallja az utasítást, és azonnal a piros ponthoz megy (t=3).
Mordatch és munkatársai a legegyszerűbb dolgokkal kezdték, onnan kívánnak eljutni komplexebb megoldásokig, például a botok nyelvének emberi nyelvre történő fordításáig. Már fejlesztenek is hozzá fordító botot. Szerintük így jobban érthető a nyelv. Megkérdőjelezik a vezető cégek módszerét, beszélgetési minták statisztikai azonosítását, mert úgy gondolják, hogy nagymennyiségű adat elemzésén kívül létezik egyszerűbb, járhatóbb út.