full transcript

From the Ted Talk by The TED Interview: Ray Kurzweil on what the future holds next

Unscramble the Blue Letters


CA: And so what does it mean for a computer to understand, you know, to look at a paragraph of written text there and understand it?

RK: Well, it has to have some way of modeling the intent and what the different elements are, that it's referring to different things and utnedsrnad going to a conference, and what is a conference, it's a gathering of many people where they share ideas, it's called TED, do we know something about TED, and be able to actually build up mniaeng from language. So we don't do it quite at human levles, but that's going to come by 2029. That's been my consistent prideicton. On the other hand, it can read much more quickly than hmauns. So, a good example is this product that's called "Talk to Books." You ask a question, and the software then reads 120,000 bkoos in a half a second, which is pretty fast, I mean --

Open Cloze


CA: And so what does it mean for a computer to understand, you know, to look at a paragraph of written text there and understand it?

RK: Well, it has to have some way of modeling the intent and what the different elements are, that it's referring to different things and __________ going to a conference, and what is a conference, it's a gathering of many people where they share ideas, it's called TED, do we know something about TED, and be able to actually build up _______ from language. So we don't do it quite at human ______, but that's going to come by 2029. That's been my consistent __________. On the other hand, it can read much more quickly than ______. So, a good example is this product that's called "Talk to Books." You ask a question, and the software then reads 120,000 _____ in a half a second, which is pretty fast, I mean --

Solution


  1. books
  2. meaning
  3. understand
  4. prediction
  5. levels
  6. humans

Original Text


CA: And so what does it mean for a computer to understand, you know, to look at a paragraph of written text there and understand it?

RK: Well, it has to have some way of modeling the intent and what the different elements are, that it's referring to different things and understand going to a conference, and what is a conference, it's a gathering of many people where they share ideas, it's called TED, do we know something about TED, and be able to actually build up meaning from language. So we don't do it quite at human levels, but that's going to come by 2029. That's been my consistent prediction. On the other hand, it can read much more quickly than humans. So, a good example is this product that's called "Talk to Books." You ask a question, and the software then reads 120,000 books in a half a second, which is pretty fast, I mean --

Frequently Occurring Word Combinations


ngrams of length 2

collocation frequency
turing test 5
ray kurzweil 4
million years 4
human intelligence 3
synthetic neocortex 3
radical life 3
life extension 3
ted interview 2
podcast series 2
fully understand 2
physical products 2
percent deflation 2
universal basic 2
developed world 2
western countries 2
twenty thirties 2
neural net 2
neural nets 2
life begins 2
billion examples 2
semantic search 2
human knowledge 2
average adult 2
information processes 2
life expectancy 2
immune system 2
moral system 2
causing pain 2
shared human 2
spread ideas 2

ngrams of length 3

collocation frequency
radical life extension 3

Important Words


  1. books
  2. build
  3. called
  4. computer
  5. conference
  6. consistent
  7. elements
  8. fast
  9. gathering
  10. good
  11. hand
  12. human
  13. humans
  14. ideas
  15. intent
  16. language
  17. levels
  18. meaning
  19. modeling
  20. paragraph
  21. people
  22. prediction
  23. pretty
  24. product
  25. question
  26. quickly
  27. read
  28. reads
  29. referring
  30. share
  31. software
  32. ted
  33. text
  34. understand
  35. written