full transcript

From the Ted Talk by Alona Fyshe: Does AI actually understand us?

Unscramble the Blue Letters


Well, I work in AI, and let me tell you, things are wild. There have been multiple examples of people being completely cnencvoid that AI understands them. In 2022, a Google engineer thought that Google’s AI was sentient. And you may have had a really human-like cnotesvoiarn with something like cptgaht. But meolds we're tiainrng today are so much better than the models we had even five years ago. It really is remarkable.

So at this super crazy moment in time, let’s ask the super crazy question: Does AI utrndansed us, or are we having our own Clever Hans moment?

Some philosophers think that computers will never understand language. To illustrate this, they developed something they call the ciehsne room agruemnt. In the Chinese room, there is a person, hypothetical perosn, who does not understand Chinese, but he has along with him a set of instructions that tell him how to respond in Chinese to any Chinese senetcne. Here's how the Chinese room wroks. A piece of paper comes in through a slot in the door, has something written in Chinese on it. The person uses their itncitonsurs to figure out how to respond. They wrtie the response down on a piece of paper and then send it back out through the door. To somebody who speaks Chinese, standing outside this room, it might seem like the person inside the room speaks Chinese. But we know they do not, because no knowledge of Chinese is required to follow the instructions. Performance on this task does not show that you know Chinese.

Open Cloze


Well, I work in AI, and let me tell you, things are wild. There have been multiple examples of people being completely _________ that AI understands them. In 2022, a Google engineer thought that Google’s AI was sentient. And you may have had a really human-like ____________ with something like _______. But ______ we're ________ today are so much better than the models we had even five years ago. It really is remarkable.

So at this super crazy moment in time, let’s ask the super crazy question: Does AI __________ us, or are we having our own Clever Hans moment?

Some philosophers think that computers will never understand language. To illustrate this, they developed something they call the _______ room ________. In the Chinese room, there is a person, hypothetical ______, who does not understand Chinese, but he has along with him a set of instructions that tell him how to respond in Chinese to any Chinese ________. Here's how the Chinese room _____. A piece of paper comes in through a slot in the door, has something written in Chinese on it. The person uses their ____________ to figure out how to respond. They _____ the response down on a piece of paper and then send it back out through the door. To somebody who speaks Chinese, standing outside this room, it might seem like the person inside the room speaks Chinese. But we know they do not, because no knowledge of Chinese is required to follow the instructions. Performance on this task does not show that you know Chinese.

Solution


  1. works
  2. convinced
  3. write
  4. understand
  5. conversation
  6. instructions
  7. chatgpt
  8. argument
  9. models
  10. chinese
  11. person
  12. sentence
  13. training

Original Text


Well, I work in AI, and let me tell you, things are wild. There have been multiple examples of people being completely convinced that AI understands them. In 2022, a Google engineer thought that Google’s AI was sentient. And you may have had a really human-like conversation with something like ChatGPT. But models we're training today are so much better than the models we had even five years ago. It really is remarkable.

So at this super crazy moment in time, let’s ask the super crazy question: Does AI understand us, or are we having our own Clever Hans moment?

Some philosophers think that computers will never understand language. To illustrate this, they developed something they call the Chinese room argument. In the Chinese room, there is a person, hypothetical person, who does not understand Chinese, but he has along with him a set of instructions that tell him how to respond in Chinese to any Chinese sentence. Here's how the Chinese room works. A piece of paper comes in through a slot in the door, has something written in Chinese on it. The person uses their instructions to figure out how to respond. They write the response down on a piece of paper and then send it back out through the door. To somebody who speaks Chinese, standing outside this room, it might seem like the person inside the room speaks Chinese. But we know they do not, because no knowledge of Chinese is required to follow the instructions. Performance on this task does not show that you know Chinese.

Frequently Occurring Word Combinations


ngrams of length 2

collocation frequency
scratch pad 18
neural network 13
understand language 5
chinese room 5
neural networks 5
network scratch 4
pad prediction 4
pretty impressive 3
clever hans 3
ai understands 3
ai understand 3
understands language 3
scratch pads 3
eighth day 2
month falls 2
super crazy 2
speaks chinese 2
brain images 2
brain image 2
prediction task 2
network scratchpad 2

ngrams of length 3

collocation frequency
neural network scratch 4
network scratch pad 4
scratch pad prediction 4
neural network scratchpad 2
ai understand language 2

Important Words


  1. ai
  2. argument
  3. call
  4. chatgpt
  5. chinese
  6. clever
  7. completely
  8. computers
  9. conversation
  10. convinced
  11. crazy
  12. developed
  13. door
  14. engineer
  15. examples
  16. figure
  17. follow
  18. google
  19. hans
  20. hypothetical
  21. illustrate
  22. instructions
  23. knowledge
  24. language
  25. models
  26. moment
  27. multiple
  28. paper
  29. people
  30. performance
  31. person
  32. philosophers
  33. piece
  34. remarkable
  35. required
  36. respond
  37. response
  38. room
  39. send
  40. sentence
  41. sentient
  42. set
  43. show
  44. slot
  45. speaks
  46. standing
  47. super
  48. task
  49. thought
  50. time
  51. today
  52. training
  53. understand
  54. understands
  55. wild
  56. work
  57. works
  58. write
  59. written
  60. years