full transcript

From the Ted Talk by Percy Liang: A new way to build AI, openly

Unscramble the Blue Letters


I was intrigued, I wtnaed to understand it, I wanted to see how far we could go with this. So I became an AI rsaherecer.

In the last 19 years, we have come a long way as a research community. Language models and more generally, foundation models, have taken off and eetnerd the mtiernaasm. But, it is important to realize that all of these achievements are based on decades of research. Research on model architectures, recsaerh on optimization algorithms, trnniaig objectives, data sets.

For a while, we had an ibnicdelre free culture, a culture of open innovation, a culture where researchers pbluesihd, researchers released data sets, code, so that others can go further. It was like a jazz ensemble where everyone was riffing off of each other, developing the technology that we have today.

Open Cloze


I was intrigued, I ______ to understand it, I wanted to see how far we could go with this. So I became an AI __________.

In the last 19 years, we have come a long way as a research community. Language models and more generally, foundation models, have taken off and _______ the __________. But, it is important to realize that all of these achievements are based on decades of research. Research on model architectures, ________ on optimization algorithms, ________ objectives, data sets.

For a while, we had an __________ free culture, a culture of open innovation, a culture where researchers _________, researchers released data sets, code, so that others can go further. It was like a jazz ensemble where everyone was riffing off of each other, developing the technology that we have today.

Solution


  1. published
  2. incredible
  3. researcher
  4. entered
  5. wanted
  6. mainstream
  7. training
  8. research

Original Text


I was intrigued, I wanted to understand it, I wanted to see how far we could go with this. So I became an AI researcher.

In the last 19 years, we have come a long way as a research community. Language models and more generally, foundation models, have taken off and entered the mainstream. But, it is important to realize that all of these achievements are based on decades of research. Research on model architectures, research on optimization algorithms, training objectives, data sets.

For a while, we had an incredible free culture, a culture of open innovation, a culture where researchers published, researchers released data sets, code, so that others can go further. It was like a jazz ensemble where everyone was riffing off of each other, developing the technology that we have today.

Frequently Occurring Word Combinations


ngrams of length 2

collocation frequency
foundation models 5
language model 3
nlp research 2
raw text 2
research community 2
training data 2
open experiment 2


Important Words


  1. achievements
  2. ai
  3. algorithms
  4. architectures
  5. based
  6. code
  7. community
  8. culture
  9. data
  10. decades
  11. developing
  12. ensemble
  13. entered
  14. foundation
  15. free
  16. generally
  17. important
  18. incredible
  19. innovation
  20. intrigued
  21. jazz
  22. language
  23. long
  24. mainstream
  25. model
  26. models
  27. objectives
  28. open
  29. optimization
  30. published
  31. realize
  32. released
  33. research
  34. researcher
  35. researchers
  36. riffing
  37. sets
  38. technology
  39. today
  40. training
  41. understand
  42. wanted
  43. years