Will Artificial Intelligence (AI) Replace Humans? — Transcript

Full transcript

1. 0:00you're in a conversation and someone
2. 0:02says artificial intelligence will soon
3. 0:04replace humans in almost everything it's
4. 0:07smarter and more efficient than us and
5. 0:09we won't be able to keep up what would
6. 0:12you say in the last 2 years machine
7. 0:14learning models and generative AI chat
8. 0:17Bots have transformed the Internet by
9. 0:19doing things that previously only humans
10. 0:21could do such as creating images
11. 0:23producing music and writing this has led
12. 0:26some to speculate that in a short time
13. 0:28artificial intelligence will overtake
14. 0:30human intelligence as a source of
15. 0:32thought and decision making for example
16. 0:35Tesla and SpaceX founder Elon Musk
17. 0:38recently said that AI will soon render
18. 0:40humans mostly obsolete predicting that
19. 0:43intelligence that is biological will be
20. 0:45less than 1% while artificial
21. 0:48intelligence has shown remarkable
22. 0:50progress Sensational claims about AI
23. 0:52replacing humans misunder understand
24. 0:54what AI is and underestimate its major
25. 0:57shortcomings they also misunderstand and
26. 1:00underestimate the uniqueness of the
27. 1:02human mind so the next time someone says
28. 1:05artificial intelligence will replace
29. 1:07humans at almost everything here are
30. 1:10three things to remember number one
31. 1:13artificial intelligence is not as smart
32. 1:15as many think at its most basic level
33. 1:18artificial intelligence isn't
34. 1:20intelligent at least not as the Mariam
35. 1:22Webster online dictionary defines it
36. 1:25intelligence is the ability to learn or
37. 1:28understand or to deal with new new or
38. 1:30trying situations except in a
39. 1:32metaphorical sense computers don't learn
40. 1:35or understand to deal with any situation
41. 1:37instead so-called machine learning
42. 1:39algorithms are instructions a computer
43. 1:42uses to process vast amounts of
44. 1:44information like images and text in
45. 1:47order to create statistical models with
46. 1:50enough input AI chat Bots like chat GPT
47. 1:53answer prompts by users at no point in
48. 1:56this process however do the transistors
49. 1:59and El ronic storage components learn
50. 2:02understand or think computers do not
51. 2:05comprehend the data they analyze the
52. 2:07questions users ask or even the answers
53. 2:10they produce they merely draw
54. 2:12conclusions from statistical patterns in
55. 2:14the data sometimes with humorous results
56. 2:18recently X's grock AI tried to summarize
57. 2:21pro basketball player Klay Thompson's
58. 2:23poor shooting performance in an NBA
59. 2:26playoff game the typically reliable
60. 2:28Golden State Warrior shot an awful 0 for
61. 2:3110 from the three-point line in a loss
62. 2:33to the Sacramento Kings pulling from
63. 2:36online fan content that teased Thompson
64. 2:38for throwing up bricks grock generated a
65. 2:41story reporting that Thompson had
66. 2:43vandalized several people's houses with
67. 2:46actual bricks the reason for the
68. 2:49misunderstanding was simple there was
69. 2:51nobody there to understand the joke even
70. 2:54if AI eventually gets better at
71. 2:56Discerning the subtleties of human humor
72. 2:58it will never laugh at the joke not
73. 3:01really because a mathematical model
74. 3:03isn't a mind no matter how well it
75. 3:05mimics one number two artificial
76. 3:09intelligence cannot create it imitates
77. 3:12and
78. 3:13extrapolates because so-called machine
79. 3:15learning is entirely dependent on the
80. 3:17training data it receives it has no
81. 3:20ability to check its models against the
82. 3:22real world everything generative AI
83. 3:25appears to create is a result of what
84. 3:27designers feed it or what's available on
85. 3:29the internet this makes chat Bots
86. 3:31especially vulnerable to error and
87. 3:34attacks for instance programmers are now
88. 3:37warning of something called Model
89. 3:39collapse when AI is trained on an
90. 3:41internet increasingly saturated with AI
91. 3:43generated material when AI models
92. 3:46consume their own output it creates a
93. 3:49feedback loop and the quality of the
94. 3:51answers they give declines Popular
95. 3:54Mechanics described an example of model
96. 3:56collapse when a team of researchers
97. 3:58trained a machine learning model on its
98. 4:00own answers for 10 Generations the AI
99. 4:04began by writing about Gothic Revival
100. 4:06architecture in Renaissance Cathedrals
101. 4:08after 10 Cycles it was babbling about
102. 4:11jack rabbits with multicolored tals of
103. 4:14course AI is also vulnerable to
104. 4:16so-called data poisoning by malicious
105. 4:19actors some warn that chat Bots could be
106. 4:21trained by strategically placed poison
107. 4:23data to spread false information about
108. 4:26public figures or steal people's
109. 4:27financial information
110. 4:30number three human intelligence is not
111. 4:32like a computer it is
112. 4:35unique for all of the apparent advances
113. 4:37in artificial intelligence we have
114. 4:39actually made very little progress
115. 4:41toward what most people Envision when
116. 4:44they think of AI a conscious adaptive
117. 4:47generally intelligent entity of the type
118. 4:50seen in science fiction this is because
119. 4:52while researchers have had great success
120. 4:54producing narrow AI like chat GPT or
121. 4:58Google's Gemini they are still mostly
122. 5:00stumped on how to create artificial
123. 5:03general intelligence or
124. 5:05AGI as John Lennox explained in his book
125. 5:082084 artificial intelligence and the
126. 5:10future of humanity narrow AI can be
127. 5:13great it's specific repetitive tasks
128. 5:16like playing chess constructing
129. 5:18sentences or identifying precancerous
130. 5:20tissue on a CAT scan however AI is not
131. 5:24great at other things that come easily
132. 5:26to humans such as navigating an
133. 5:29unfamiliar room or detecting sarcasm
134. 5:32this is because so far AI lacks the kind
135. 5:35of generalized intelligence that allows
136. 5:37us to move from task to task to think in
137. 5:40the abstract to apply background
138. 5:43knowledge to use common sense and to
139. 5:45understand cause and effect for all the
140. 5:48hype around machine learning AI systems
141. 5:51continue to be at the most fundamental
142. 5:53level programs that do what their
143. 5:55creators tell them to do the human mind
144. 5:59in other words is unlike anything the
145. 6:01world of computing has yet produced this
146. 6:03is partly because as psychologist Robert
147. 6:05Epstein argued several years ago at Aon
148. 6:08the human brain is not a computer it
149. 6:12does not process information retrieve
150. 6:14knowledge or store memories as symbolic
151. 6:17data these he explained are metaphors we
152. 6:21have come to use in the digital age just
153. 6:23as people in past ages likened the mind
154. 6:25to a steam engine a mechanical clock or
155. 6:28a set of pipes
156. 6:30we use such technological metaphors
157. 6:32because we don't really understand
158. 6:33ourselves for instance scientists still
159. 6:36do not know where if anywhere in the
160. 6:39brain human consciousness resides how it
161. 6:42emerges or even what it is and the
162. 6:45concept of the soul is unexplained by
163. 6:48science the idea that we would ever be
164. 6:50able to create a computer that Rivals
165. 6:52human intelligence or as conscious
166. 6:55intentional or truly creative as we are
167. 6:58is much more far-fetched than our
168. 7:00science fiction saturated culture would
169. 7:02lead you to believe AI will almost
170. 7:05certainly become a major part of life in
171. 7:07the coming years and it will replace
172. 7:09humans in many tasks that rely on
173. 7:11repetitive predictable skills but it is
174. 7:14not in the most meaningful sense smarter
175. 7:17than we are nor is it likely to render
176. 7:19the human mind obsolete so the next time
177. 7:22someone says artificial intelligence
178. 7:24will replace humans at almost everything
179. 7:27remember these three things number
180. 7:29number one artificial intelligence is
181. 7:31not as smart as many think number two
182. 7:34artificial intelligence cannot create it
183. 7:38imitates and
184. 7:39extrapolates and number three human
185. 7:42intelligence is not like a computer it's
186. 7:45Unique