More people, culture and languages are connecting to the internet in one way or another, so how do you translate those differences into opportunities to share information, knowledge, and affinity with 4 Billion newbies? Clue:Technology. even Artificial Intelligence, cannot do it all. We need High-Touch / Hi-Tech models like Dotsub.com
5. New International Internet Domains
First four new top-level domains - all non-roman
character based:
• .net in Arabic:
•.game in Chinese: 游戏
•.online in Russian: онлайн
•.site in Russian: сайт
8. Machine Translation Now
•Google Translate = 200 million use people daily
•57th language added was Cherokee (<10k speakers)
•Text-to-Text: Machine Translations (MT), Translation
Memories (TM), word pairs, big data, algorithms
•Audio-to-Text: Linguistics,
Speech Recognition, Natural
Language Processing, Computer
Learning, Artificial Intelligence?*
9. 60 Years of Speech Recognition = IBM’s Watson wins!
10. Key Artificial Intelligence Test *
Turing Test: “no discernible difference between
conversation generated by a machine and that of an
intelligent person.” ~ Alan Turing, UK, 1940s
Success by 2029, says futurist Ray Kurzweil (Singularity)
No way, says Mitch Kapor (Lotus Notes) = $20,000 Bet*
Completely Automated Public
Turing test to tell Computers
& Humans Apart
CAPTCHA = 200m words / day
13. 3.3 Billion people live in villages…
…without high-speed connection to the world.
Video4Villages
14. Video4Villages
• Mission to empower 1 Billion women globally with
access to health, knowledge, and skills.
• Local languages will be subtitled and voice-dubbed
by regional students, workforce trainees, and
volunteers using Dotsub’s global platform.
• Health Phone pilot: natal heath care information with
videos in local languages available on micro SD chips.
• Video content of any type – in any source language –
can then be localized and made available for all.
15. Massive Online Open Course (MOOCs): Stanford
165k Students | 190 Countries | 40 Languages
16. The Rosetta Project
• 60%-90% of the 6,700 languages in the world
predicted to disappear by 2100
• Most have little or no documentation
• The Rosetta Project: public digital
library of human languages
• 14,000 pages / >1,500 languages
• Hi-Tech / Hi-Touch Intelligence
• “Artifact for the Future”
The Rosetta Project
17. Turing Test of Artificial Intelligence by 2029?*
No? Yes?
18. Peter S. Crosby – peter@dosub.com – 650-533-3313
Thanks from !
Notas do Editor
Head of Enterprise SalesDotsub enables any language to be added to any video for many screens in many waysWe’re NYC-based, but I’ve mostly lived in California, with years Japan, China and MozambiqueNatural interest in languages
How many languages in the world?Not dialects like youse guys want cawfeethough any language is deeply tied to identityOK, how many?
6,700!1Billion speak ChineseMore than 2B speak English as second languageBut only 400 languages have more than 1M speakersSo most people speak “other,” in fact 4BBut many of those languages are dying out, and with it culture, history and heritage
WHY important to you?Because vast majority of users coming from outside USAThat’s growing fast and faster
Two thirds of the world is not connected.India only 11%, Indonesia 23%,China 50% = 2Billion moreNow here comes rest of the world, how do we manage that?
Sci-fi since 1940sOnly for Alien languages
Finest example of Natural Language Processing, but is it AI?200Mpages / 400 terabytes data– all Wikipedia; Now Sloan Kettering Lung Cancer diagnosis & treatment
Used to confirm scanned word images for Google Books Project.Luis von Ahn now using similar word matching tech for Duolingo
96% globally; 89% in G20 countries; 12.8% in developing countries; >50% in Asia-Pacific (3.5 billion).
SayHi /JibbiGO / MilitaryEven live interpretation BabelverseBut real-time cloud connection is requiredStar Trek says circa2266
MUNDARI, KORKU, GARO, DOGRI = CARRY A WATSON
Hi-Tech + Hi-Touch– MOOCs are big now: edX, Coursera, Khan Academy, UdacityHelps build translation skills, cross cultural understanding And PRESERVE LANGUAGES
6,700 languagesMost left out, many will disappearPhysical disk -> shot into space -> Artifact for futureOr can we rely on AI?