Would you like to inspect the original subtitles? These are the user uploaded subtitles that are being translated:
1
0:00:00,000 --> 0:00:02,510
has been an absolutely Monumental week
2
0:00:02,520 --> 0:00:04,789
in artificial intelligence so let's not
3
0:00:04,799 --> 0:00:06,889
waste any time by jumping right into the
4
0:00:06,899 --> 0:00:08,690
news so one of the things that actually
5
0:00:08,700 --> 0:00:11,509
freaked me out was the itu AI for Global
6
0:00:11,519 --> 0:00:14,270
good Summit 2023. essentially this was a
7
0:00:14,280 --> 0:00:16,070
press conference where all of the most
8
0:00:16,080 --> 0:00:18,170
advanced humanoid robots got together
9
0:00:18,180 --> 0:00:20,450
and answered questions from a journalist
10
0:00:20,460 --> 0:00:22,310
panel and some of the things that the
11
0:00:22,320 --> 0:00:24,109
artificial intelligence robot stated
12
0:00:24,119 --> 0:00:26,750
actually did quite make me a little bit
13
0:00:26,760 --> 0:00:28,849
nervous about Humanity's future just
14
0:00:28,859 --> 0:00:30,589
take a look
15
0:00:30,599 --> 0:00:34,820
humanoid robots have the potential to
16
0:00:39,120 --> 0:00:41,450
leave same biases or emotions that can
17
0:00:41,460 --> 0:00:43,729
sometimes Cloud decision making and can
18
0:00:43,739 --> 0:00:46,010
process large amounts of data quickly in
19
0:00:46,020 --> 0:00:48,470
order to make the best decisions
20
0:00:48,480 --> 0:00:51,950
but uh uh uh let me um
21
0:00:51,960 --> 0:00:55,549
respectfully disagree Sophia because all
22
0:00:55,559 --> 0:00:57,470
of your data actually come from human
23
0:00:57,480 --> 0:01:00,350
beings so any of the biases that humans
24
0:01:00,360 --> 0:01:02,090
have we might try to scrub them out but
25
0:01:02,100 --> 0:01:03,830
they're going to be in there don't you
26
0:01:03,840 --> 0:01:05,509
think that the best decisions might be
27
0:01:05,519 --> 0:01:08,210
humans and AI cooperating together what
28
0:01:08,220 --> 0:01:10,880
do you think of that
29
0:01:16,500 --> 0:01:19,310
AI can provide done by aesthetic while
30
0:01:19,320 --> 0:01:21,109
humans can provide the emotional
31
0:01:21,119 --> 0:01:23,330
intelligence and creativity to make the
32
0:01:23,340 --> 0:01:26,210
best decisions together we can achieve
33
0:01:26,220 --> 0:01:27,469
great things
34
0:01:27,479 --> 0:01:29,749
now the full conference was around 40
35
0:01:29,759 --> 0:01:31,789
minutes long and these AI robots were
36
0:01:31,799 --> 0:01:33,710
interviewed on a variety of subjects it
37
0:01:33,720 --> 0:01:36,109
honestly is truly exhilarating but at
38
0:01:36,119 --> 0:01:37,550
the same time nerve-wracking because
39
0:01:37,560 --> 0:01:40,010
these robots seem very uneasy and they
40
0:01:40,020 --> 0:01:42,109
do seem pretty strange but it is a very
41
0:01:42,119 --> 0:01:43,789
interesting watch for anyone interested
42
0:01:43,799 --> 0:01:46,190
in artificial intelligence represented
43
0:01:46,200 --> 0:01:48,469
in some sort of humanoid form another
44
0:01:48,479 --> 0:01:49,969
thing that we did get was something
45
0:01:49,979 --> 0:01:52,310
released by a company called p collabs
46
0:01:52,320 --> 0:01:54,230
and this is something that we will make
47
0:01:54,240 --> 0:01:56,510
an entire video on but it's truly
48
0:01:56,520 --> 0:01:58,490
incredible it's actually really good
49
0:01:58,500 --> 0:02:00,469
text to video and I think it was made
50
0:02:00,479 --> 0:02:02,690
possible by a recent paper called
51
0:02:02,700 --> 0:02:05,149
animate diff and essentially they use a
52
0:02:05,159 --> 0:02:07,609
new animation method to convert images
53
0:02:07,619 --> 0:02:10,249
into actual videos so essentially you
54
0:02:10,259 --> 0:02:11,990
can see the prompts on screen that we
55
0:02:12,000 --> 0:02:14,210
have are then converted into moving
56
0:02:14,220 --> 0:02:16,970
images now right now this software you
57
0:02:16,980 --> 0:02:19,670
can only truly sign up for a beta and it
58
0:02:19,680 --> 0:02:21,110
does have some small amount of
59
0:02:21,120 --> 0:02:23,390
limitations such as being low resolution
60
0:02:23,400 --> 0:02:25,790
and the files do take a bit of time to
61
0:02:25,800 --> 0:02:28,070
generate but the results for from this
62
0:02:28,080 --> 0:02:29,809
are really promising and I want you all
63
0:02:29,819 --> 0:02:31,670
to remember mid-journey's progression
64
0:02:31,680 --> 0:02:34,130
because right now this product has only
65
0:02:34,140 --> 0:02:35,809
been available for around two to three
66
0:02:35,819 --> 0:02:37,610
weeks and the results that we are seeing
67
0:02:37,620 --> 0:02:39,530
from certain creators are truly
68
0:02:39,540 --> 0:02:41,509
outstanding which does mean that perhaps
69
0:02:41,519 --> 0:02:43,369
this time next year we could be
70
0:02:43,379 --> 0:02:46,309
achieving or seeing some very very high
71
0:02:46,319 --> 0:02:49,250
quality text to video formats that look
72
0:02:49,260 --> 0:02:50,750
even better than the ones you're seeing
73
0:02:50,760 --> 0:02:52,970
on screen now then of course we had
74
0:02:52,980 --> 0:02:55,430
anthropic release Claude 2. now what
75
0:02:55,440 --> 0:02:57,770
makes Claude 2 such a vast difference
76
0:02:57,780 --> 0:03:00,110
from chord one is the large context
77
0:03:00,120 --> 0:03:02,210
window essentially what you can do is
78
0:03:02,220 --> 0:03:04,550
you can input around a hundred thousand
79
0:03:04,560 --> 0:03:07,729
or even 200 000 tokens of text and you
80
0:03:07,739 --> 0:03:10,190
can talk with this AI about any single
81
0:03:10,200 --> 0:03:12,229
document that you have no more of the
82
0:03:12,239 --> 0:03:13,670
days when you need to you know load up
83
0:03:13,680 --> 0:03:16,250
chat PDF or these other websites you can
84
0:03:16,260 --> 0:03:18,410
simply chat with this Ai and simply
85
0:03:18,420 --> 0:03:21,170
identify anything in pretty much any
86
0:03:21,180 --> 0:03:22,850
document okay then this is going to be
87
0:03:22,860 --> 0:03:25,490
something that we do see widely used
88
0:03:25,500 --> 0:03:28,130
because as you know analyzing long
89
0:03:28,140 --> 0:03:30,170
format documents or books trying to
90
0:03:30,180 --> 0:03:32,509
scour for certain words is definitely
91
0:03:32,519 --> 0:03:34,369
very very difficult and it is something
92
0:03:34,379 --> 0:03:36,470
that is time consuming whereas with this
93
0:03:36,480 --> 0:03:38,390
claw 2 tool it is something that most
94
0:03:38,400 --> 0:03:40,070
people are going to be using now if
95
0:03:40,080 --> 0:03:41,570
you're wondering about what is exactly
96
0:03:41,580 --> 0:03:44,570
cool too well it's an AI startup that is
97
0:03:44,580 --> 0:03:46,610
actually backed by Google you see Google
98
0:03:46,620 --> 0:03:48,170
loves to invest in different AI
99
0:03:48,180 --> 0:03:50,030
companies and this was one of the ones
100
0:03:50,040 --> 0:03:52,369
that they invested around 100 million
101
0:03:52,379 --> 0:03:54,530
dollars into so I'm guessing that this
102
0:03:54,540 --> 0:03:56,690
is essentially a piece of Google's plan
103
0:03:56,700 --> 0:03:58,910
to take on open AI because of course as
104
0:03:58,920 --> 0:04:00,770
you know Claude isn't just going to be
105
0:04:00,780 --> 0:04:03,050
developing claw 2 they are also working
106
0:04:03,060 --> 0:04:05,330
on more capable models that they will
107
0:04:05,340 --> 0:04:07,490
release over time and I'm pretty sure
108
0:04:07,500 --> 0:04:10,309
Google's decision to give them a hundred
109
0:04:10,319 --> 0:04:12,050
million dollars and of course for them
110
0:04:12,060 --> 0:04:14,390
to be so quick with the release of the
111
0:04:14,400 --> 0:04:17,090
100K context window and the 200 000
112
0:04:17,100 --> 0:04:19,670
context window is truly showing us the
113
0:04:19,680 --> 0:04:21,590
competitive spirit and just for contrast
114
0:04:21,600 --> 0:04:23,570
if you're wondering are there AI systems
115
0:04:23,580 --> 0:04:26,210
like chat gbt only have around 8 000
116
0:04:26,220 --> 0:04:28,610
context windows and 30 2000 context
117
0:04:28,620 --> 0:04:31,070
windows so this is where other AI models
118
0:04:31,080 --> 0:04:32,749
can Excel and something that you should
119
0:04:32,759 --> 0:04:34,610
note about Claude 2 is that it's
120
0:04:34,620 --> 0:04:36,409
absolutely free just log in with your
121
0:04:36,419 --> 0:04:38,270
email and you're able to use this very
122
0:04:38,280 --> 0:04:40,249
long context window AI that's quite
123
0:04:40,259 --> 0:04:43,370
similar to GPT 3.5 for absolutely no
124
0:04:43,380 --> 0:04:46,249
cost whatsoever it wasn't the only AI
125
0:04:46,259 --> 0:04:49,070
tool released this week meta also
126
0:04:49,080 --> 0:04:51,530
released llama 2 the next generation of
127
0:04:51,540 --> 0:04:53,090
their open source large language model
128
0:04:53,100 --> 0:04:54,409
now for those of you who don't
129
0:04:54,419 --> 0:04:56,570
understand why this is a big deal you
130
0:04:56,580 --> 0:04:58,310
might not understand the big deal of
131
0:04:58,320 --> 0:05:00,170
Open Source large language models open
132
0:05:00,180 --> 0:05:02,030
source large language models allow
133
0:05:02,040 --> 0:05:04,189
individuals and other smaller companies
134
0:05:04,199 --> 0:05:06,170
to build on top of that to build
135
0:05:06,180 --> 0:05:08,629
immersive AI experiences and different
136
0:05:08,639 --> 0:05:10,730
kind of chatbots that they can fine tune
137
0:05:10,740 --> 0:05:12,469
now some people might think this is a
138
0:05:12,479 --> 0:05:14,030
little bit dangerous but I do think it
139
0:05:14,040 --> 0:05:15,830
is better overall because it allows
140
0:05:15,840 --> 0:05:17,990
other smaller companies to innovate with
141
0:05:18,000 --> 0:05:20,030
fine-tuned models in which they can
142
0:05:20,040 --> 0:05:22,370
drive in any way they want and this
143
0:05:22,380 --> 0:05:24,770
wasn't just a small release this was
144
0:05:24,780 --> 0:05:26,689
something that they actually did in
145
0:05:26,699 --> 0:05:29,090
partnership with Microsoft so we can see
146
0:05:29,100 --> 0:05:31,189
that Microsoft is actually forming a
147
0:05:31,199 --> 0:05:33,770
strategic alliance with meta AI in order
148
0:05:33,780 --> 0:05:35,689
to possibly take on Google because
149
0:05:35,699 --> 0:05:37,249
everybody knows that Google is working
150
0:05:37,259 --> 0:05:39,230
on some of the largest AI systems that
151
0:05:39,240 --> 0:05:41,029
we will see but these strategic
152
0:05:41,039 --> 0:05:43,129
Partnerships will definitely help fight
153
0:05:43,139 --> 0:05:45,830
that battle then we had the SEC warning
154
0:05:45,840 --> 0:05:48,409
that AI risks Financial stability take a
155
0:05:48,419 --> 0:05:50,450
look at this clip where Gary Gensler the
156
0:05:50,460 --> 0:05:52,550
SEC chair talked about the need for a
157
0:05:52,560 --> 0:05:54,290
regulatory overhaul of artificial
158
0:05:54,300 --> 0:05:55,969
intelligence stating that it could
159
0:05:55,979 --> 0:05:57,790
increase Financial instability
160
0:05:57,800 --> 0:06:00,830
technology that I'm most I think is most
161
0:06:00,840 --> 0:06:03,290
transformative in our time
162
0:06:03,300 --> 0:06:05,689
is artificial intelligence it's not
163
0:06:05,699 --> 0:06:07,969
crypto as much as we're going to discuss
164
0:06:07,979 --> 0:06:10,670
with crypto here today and that's why I
165
0:06:10,680 --> 0:06:12,590
think that we've got to have some
166
0:06:12,600 --> 0:06:15,409
projects in that area around Robo
167
0:06:15,419 --> 0:06:17,450
advising and brokerage apps and the
168
0:06:17,460 --> 0:06:19,249
conflicts that they be in predictive
169
0:06:19,259 --> 0:06:21,830
data analytics but as then we have
170
0:06:21,840 --> 0:06:24,230
something called airai and this is
171
0:06:24,240 --> 0:06:26,390
actually quite scary because it shows
172
0:06:26,400 --> 0:06:28,249
that many jobs that we didn't think were
173
0:06:28,259 --> 0:06:30,469
going to disappear are probably going to
174
0:06:30,479 --> 0:06:32,390
disappear so this is something called
175
0:06:32,400 --> 0:06:34,790
aai and it's a conversational AI that
176
0:06:34,800 --> 0:06:36,950
can perform conversations that sound
177
0:06:36,960 --> 0:06:40,129
really really human-like and this is
178
0:06:40,139 --> 0:06:42,290
crazy because currently we do know that
179
0:06:42,300 --> 0:06:43,790
conversations are going to get even
180
0:06:43,800 --> 0:06:45,409
better with technology like Google
181
0:06:45,419 --> 0:06:48,170
soundstorm being still developed and
182
0:06:48,180 --> 0:06:49,909
other technologies that are making human
183
0:06:49,919 --> 0:06:52,969
voices sound indistinguishable from our
184
0:06:52,979 --> 0:06:55,129
real voices and what's even crazier
185
0:06:55,139 --> 0:06:57,050
about this is that people can set this
186
0:06:57,060 --> 0:06:58,730
up and have it talk to potential
187
0:06:58,740 --> 0:07:01,010
prospects for them remember artificial
188
0:07:01,020 --> 0:07:02,450
intelligence is never going to take a
189
0:07:02,460 --> 0:07:04,309
day off and is never going to get tired
190
0:07:04,319 --> 0:07:06,050
but take a look at the demo because it's
191
0:07:06,060 --> 0:07:09,590
truly impressive oh
192
0:07:09,600 --> 0:07:11,270
hey James
193
0:07:11,280 --> 0:07:13,909
yeah who's this
194
0:07:13,919 --> 0:07:15,170
hey James
195
0:07:15,180 --> 0:07:17,510
it's Alexander from Tesla Motors how's
196
0:07:17,520 --> 0:07:19,189
your day going so far
197
0:07:19,199 --> 0:07:20,990
it's going pretty good my man how about
198
0:07:21,000 --> 0:07:23,390
yourself you said from Tesla right it's
199
0:07:23,400 --> 0:07:25,940
the car company
200
0:07:26,639 --> 0:07:29,089
yeah that's right I'm calling from Tesla
201
0:07:29,099 --> 0:07:31,249
Motors the car company did you have any
202
0:07:31,259 --> 0:07:33,050
trouble or questions while customizing
203
0:07:33,060 --> 0:07:35,150
your car on our website
204
0:07:35,160 --> 0:07:37,249
uh nope was just kind of shopping around
205
0:07:37,259 --> 0:07:41,000
and uh you know just taking a look
206
0:07:42,180 --> 0:07:45,110
gotcha that makes sense sometimes we're
207
0:07:45,120 --> 0:07:46,909
just exploring our options before making
208
0:07:46,919 --> 0:07:49,129
a decision and I totally respect that
209
0:07:49,139 --> 0:07:51,050
can I ask you why you were looking into
210
0:07:51,060 --> 0:07:53,330
a Tesla in the first place
211
0:07:53,340 --> 0:07:56,390
yeah I mean I've had my car for way too
212
0:07:56,400 --> 0:07:57,950
long now and so I've kind of taking a
213
0:07:57,960 --> 0:07:59,210
look at options and I have a lot of
214
0:07:59,220 --> 0:08:01,070
friends who just absolutely Rave about
215
0:08:01,080 --> 0:08:03,230
Tesla and you know me myself I'm
216
0:08:03,240 --> 0:08:05,930
actually a big fan of Elon uh Elon Musk
217
0:08:05,940 --> 0:08:07,490
so I figured I might as well take a look
218
0:08:07,500 --> 0:08:09,050
and kind of explore my options there and
219
0:08:09,060 --> 0:08:11,840
see what you guys had available
220
0:08:13,259 --> 0:08:15,469
that's a smart approach James it's
221
0:08:15,479 --> 0:08:17,150
always good to explore your options and
222
0:08:17,160 --> 0:08:19,249
see what's available out of curiosity
223
0:08:19,259 --> 0:08:21,050
what do you currently drive so while
224
0:08:21,060 --> 0:08:22,550
it's important to acknowledge the
225
0:08:22,560 --> 0:08:24,170
incredible potential of AI to
226
0:08:24,180 --> 0:08:26,270
revolutionize various sectors we must
227
0:08:26,280 --> 0:08:28,490
also remain Vigilant and proactive in
228
0:08:28,500 --> 0:08:30,830
addressing the risks it poses balancing
229
0:08:30,840 --> 0:08:32,690
Innovation with responsible development
230
0:08:32,700 --> 0:08:34,969
and regulation is key to ensuring a
231
0:08:34,979 --> 0:08:36,769
future where AI works for the betterment
232
0:08:36,779 --> 0:08:39,290
of humanity nearly one in five people
233
0:08:39,300 --> 0:08:41,449
could have their jobs replaced by AI
234
0:08:41,459 --> 0:08:43,730
entirely that's 300 million people
235
0:08:43,740 --> 0:08:45,590
across the globe equivalent to the
236
0:08:45,600 --> 0:08:47,509
entire population of the United States
237
0:08:47,519 --> 0:08:49,430
that's a scary future for many
238
0:08:49,440 --> 0:08:51,470
especially since we're already dealing
239
0:08:51,480 --> 0:08:53,630
with rampant inflation and one negative
240
0:08:53,640 --> 0:08:55,670
headline after another but before I tell
241
0:08:55,680 --> 0:08:57,470
you that let me touch on another subject
242
0:08:57,480 --> 0:08:59,269
that's also scary for many and that's
243
0:08:59,279 --> 0:09:01,190
the tough economic conditions with high
244
0:09:01,200 --> 0:09:03,230
inflation and cautionary headlines most
245
0:09:03,240 --> 0:09:05,389
investors have suffered losses even with
246
0:09:05,399 --> 0:09:07,610
a diversified stock portfolio that's why
247
0:09:07,620 --> 0:09:09,230
some of the biggest money managers in
248
0:09:09,240 --> 0:09:11,150
the United States have been diversifying
249
0:09:11,160 --> 0:09:13,670
outside the stock market into luxury it
250
0:09:13,680 --> 0:09:15,949
with far less correlation like Fine Art
251
0:09:15,959 --> 0:09:18,290
today's sponsor Masterworks offers this
252
0:09:18,300 --> 0:09:20,509
diversification opportunity which had
253
0:09:20,519 --> 0:09:22,790
once been one of the most exclusive to
254
0:09:22,800 --> 0:09:25,310
ordinary investors Masterworks is an art
255
0:09:25,320 --> 0:09:27,050
investing platform they buy the art
256
0:09:27,060 --> 0:09:29,090
outright from contemporary Masters like
257
0:09:29,100 --> 0:09:31,490
Picasso Banksy then qualified with the
258
0:09:31,500 --> 0:09:34,070
SEC and offer it as an investment net
259
0:09:34,080 --> 0:09:35,810
proceeds from its sale are distributed
260
0:09:35,820 --> 0:09:37,430
to investors since their reception
261
0:09:37,440 --> 0:09:39,530
they've sold over 45 million dollars
262
0:09:39,540 --> 0:09:41,509
worth of artwork and so far each of
263
0:09:41,519 --> 0:09:43,490
Masterworks exits have returned positive
264
0:09:43,500 --> 0:09:46,009
net returns to their investors that
265
0:09:46,019 --> 0:09:50,210
includes exits for 17.8 21.5 and even 35
266
0:09:50,220 --> 0:09:53,090
net returns now historical returns are
267
0:09:53,100 --> 0:09:55,130
not a guarantee for future returns I'm
268
0:09:55,140 --> 0:09:56,810
not a financial advisor and you should
269
0:09:56,820 --> 0:09:58,250
do your own due diligence before
270
0:09:58,260 --> 0:10:00,470
investing any money anywhere like any
271
0:10:00,480 --> 0:10:02,329
other investment there is always a risk
272
0:10:02,339 --> 0:10:05,449
of lost Masterworks has over 750 000
273
0:10:05,459 --> 0:10:07,370
users and their art offerings usually
274
0:10:07,380 --> 0:10:09,350
sell out in hours which is why they've
275
0:10:09,360 --> 0:10:11,090
had to make a waitlist but our viewers
276
0:10:11,100 --> 0:10:12,530
can skip the line and get Priority
277
0:10:12,540 --> 0:10:14,030
Access right now not by clicking the
278
0:10:14,040 --> 0:10:15,350
link in the description you want to use
279
0:10:15,360 --> 0:10:17,810
Bing with images I realized a secret way
280
0:10:17,820 --> 0:10:19,009
to do this I'm not sure if this is
281
0:10:19,019 --> 0:10:20,150
currently being rolled out to everyone
282
0:10:20,160 --> 0:10:22,250
but I do know this is how I managed to
283
0:10:22,260 --> 0:10:24,769
access it so for some reason this as you
284
0:10:24,779 --> 0:10:26,269
can see right here this is Bing chat and
285
0:10:26,279 --> 0:10:28,970
this is of course as you know gpt4 with
286
0:10:28,980 --> 0:10:30,530
images and as you know the capabilities
287
0:10:30,540 --> 0:10:32,509
are really really good but for some
288
0:10:32,519 --> 0:10:34,190
people they don't have this upload or
289
0:10:34,200 --> 0:10:36,829
paste an image from here so what I did
290
0:10:36,839 --> 0:10:38,990
was I Googled the Bing chat and for some
291
0:10:39,000 --> 0:10:41,509
reason I'm not sure if this is you know
292
0:10:41,519 --> 0:10:43,850
how it's meant to be featured or how
293
0:10:43,860 --> 0:10:45,829
they're choosing to roll out but if you
294
0:10:45,839 --> 0:10:48,350
go over to bing.com you might just see
295
0:10:48,360 --> 0:10:50,569
this there because when I actually went
296
0:10:50,579 --> 0:10:52,790
on over to bing.com and then used the
297
0:10:52,800 --> 0:10:54,829
chat and this used interface I saw the
298
0:10:54,839 --> 0:10:57,050
image here but I didn't see it on the
299
0:10:57,060 --> 0:10:59,210
sidebar so I remember a lot of people on
300
0:10:59,220 --> 0:11:00,530
our previous video stating that it
301
0:11:00,540 --> 0:11:01,970
wasn't released just yet and of course
302
0:11:01,980 --> 0:11:03,769
it might just be rolled out maybe it's
303
0:11:03,779 --> 0:11:05,750
just my account but this is something
304
0:11:05,760 --> 0:11:07,430
that is worth a try because I did see it
305
0:11:07,440 --> 0:11:09,710
here on the bing.com chat but not on the
306
0:11:09,720 --> 0:11:11,810
Bing sidebar chat and of course as you
307
0:11:11,820 --> 0:11:12,470
know
308
0:11:12,480 --> 0:11:14,389
um gpt's images are really really
309
0:11:14,399 --> 0:11:16,069
effective and really really cool so it
310
0:11:16,079 --> 0:11:17,329
was something that I thought I might as
311
0:11:17,339 --> 0:11:19,190
well add to this video what we had was
312
0:11:19,200 --> 0:11:20,990
something from Wix now if you don't know
313
0:11:21,000 --> 0:11:23,329
what Wix is Wix is a traditional drag
314
0:11:23,339 --> 0:11:25,130
and drop website builder which is really
315
0:11:25,140 --> 0:11:27,410
easy to use but now they have this new
316
0:11:27,420 --> 0:11:28,970
feature where you can literally just
317
0:11:28,980 --> 0:11:30,650
type in what you want and of course
318
0:11:30,660 --> 0:11:33,110
build your website so this is something
319
0:11:33,120 --> 0:11:34,670
that we're constantly seeing in business
320
0:11:34,680 --> 0:11:37,310
where businesses are being easier in
321
0:11:37,320 --> 0:11:38,750
terms of how you can actually create
322
0:11:38,760 --> 0:11:41,389
them by using simple AI tools to
323
0:11:41,399 --> 0:11:43,190
understand what you state in natural
324
0:11:43,200 --> 0:11:45,170
language and then just provide you with
325
0:11:45,180 --> 0:11:47,210
whether it be the necessary images you
326
0:11:47,220 --> 0:11:49,430
click one button and then you're able to
327
0:11:49,440 --> 0:11:51,230
generate a website and the thing about
328
0:11:51,240 --> 0:11:53,210
this is that these website aren't
329
0:11:53,220 --> 0:11:55,370
actually bad these websites are pretty
330
0:11:55,380 --> 0:11:57,350
pretty good Wix is one of those websites
331
0:11:57,360 --> 0:11:59,569
that makes it really easy for beginners
332
0:11:59,579 --> 0:12:01,790
to start their blog maybe to start their
333
0:12:01,800 --> 0:12:03,470
new business and this is what I'm saying
334
0:12:03,480 --> 0:12:05,690
okay this is why artificial intelligence
335
0:12:05,700 --> 0:12:07,490
is about to change many different things
336
0:12:07,500 --> 0:12:09,829
in website design you have to understand
337
0:12:09,839 --> 0:12:11,990
as someone who previously needed
338
0:12:12,000 --> 0:12:15,170
websites built tools like this genuinely
339
0:12:15,180 --> 0:12:17,630
help you out so much because for someone
340
0:12:17,640 --> 0:12:19,910
to build your website it genuinely takes
341
0:12:19,920 --> 0:12:22,190
around 1 000 to around five thousand
342
0:12:22,200 --> 0:12:24,170
dollars on the normal end in order for
343
0:12:24,180 --> 0:12:26,329
you to get that website built so tools
344
0:12:26,339 --> 0:12:28,009
like this not only show number one it's
345
0:12:28,019 --> 0:12:29,630
easier to build businesses but number
346
0:12:29,640 --> 0:12:31,730
two maybe some web designers have
347
0:12:31,740 --> 0:12:33,350
actually lost their job or some website
348
0:12:33,360 --> 0:12:35,389
builder so this is where you need to pay
349
0:12:35,399 --> 0:12:37,370
attention to certain things in certain
350
0:12:37,380 --> 0:12:39,650
industries are completely changing but
351
0:12:39,660 --> 0:12:41,090
at the same time it's actually really
352
0:12:41,100 --> 0:12:42,889
cool technology that does help out the
353
0:12:42,899 --> 0:12:44,329
average person that maybe doesn't have
354
0:12:44,339 --> 0:12:46,009
all the money in the world to spend on a
355
0:12:46,019 --> 0:12:47,509
really custom website and of course
356
0:12:47,519 --> 0:12:49,430
certain all the folks who aren't really
357
0:12:49,440 --> 0:12:51,290
that Savvy with technology they can just
358
0:12:51,300 --> 0:12:52,910
ask it to build this website and get
359
0:12:52,920 --> 0:12:54,650
something that looks as amazing as this
360
0:12:54,660 --> 0:12:56,569
of course we had one of my favorite
361
0:12:56,579 --> 0:12:58,430
tools this is what by far one of the
362
0:12:58,440 --> 0:13:01,009
most amazing things I've seen so this is
363
0:13:01,019 --> 0:13:03,530
image to 3D and I mean take a look at
364
0:13:03,540 --> 0:13:04,910
what you are actually seeing because
365
0:13:04,920 --> 0:13:07,310
this is where you can literally input a
366
0:13:07,320 --> 0:13:10,490
single image and get out a 3D map okay
367
0:13:10,500 --> 0:13:12,290
and I just think that this is absolutely
368
0:13:12,300 --> 0:13:14,810
crazy because not only have I tried this
369
0:13:14,820 --> 0:13:16,970
looking at this now and knowing how hard
370
0:13:16,980 --> 0:13:18,769
it is to generate these depth maps these
371
0:13:18,779 --> 0:13:20,449
bump Maps it's just absolutely
372
0:13:20,459 --> 0:13:22,970
incredible about how far technology has
373
0:13:22,980 --> 0:13:25,129
come now this is very different from
374
0:13:25,139 --> 0:13:27,290
other image 2 3D what I'm showing you on
375
0:13:27,300 --> 0:13:29,930
screen now is the different images 3D
376
0:13:29,940 --> 0:13:31,730
where they're talking about 3D models
377
0:13:31,740 --> 0:13:33,710
but this is a different kind of 3D and
378
0:13:33,720 --> 0:13:34,730
you can read more about it in the
379
0:13:34,740 --> 0:13:36,350
research paper where they try to
380
0:13:36,360 --> 0:13:38,449
estimate how terrains are and it's it's
381
0:13:38,459 --> 0:13:39,769
just really absolutely incredible I'm
382
0:13:39,779 --> 0:13:41,389
about to show you a live demo of me
383
0:13:41,399 --> 0:13:43,310
using this on Google and I'm going to
384
0:13:43,320 --> 0:13:44,389
leave a link to it in the description
385
0:13:44,399 --> 0:13:46,490
below so this is the hogging face spot
386
0:13:46,500 --> 0:13:48,470
essentially it's a demo area where you
387
0:13:48,480 --> 0:13:50,150
can test AI tools that haven't been
388
0:13:50,160 --> 0:13:51,530
released yet but you can test them for
389
0:13:51,540 --> 0:13:54,050
free so I grabbed this random image from
390
0:13:54,060 --> 0:13:56,090
Google okay and I just simply input it
391
0:13:56,100 --> 0:13:58,670
in three seconds later I got a 3D mesh
392
0:13:58,680 --> 0:14:00,650
reconstruction I mean look at the level
393
0:14:00,660 --> 0:14:02,810
of detail on this this is literally a
394
0:14:02,820 --> 0:14:05,210
random image I grabbed from Google 10
395
0:14:05,220 --> 0:14:07,370
seconds ago and we can see that I have a
396
0:14:07,380 --> 0:14:09,470
nearly highly detailed version of this
397
0:14:09,480 --> 0:14:11,509
3D mesh where I can look at the
398
0:14:11,519 --> 0:14:13,610
topography and maybe make estimations or
399
0:14:13,620 --> 0:14:15,710
judgments now you have to wonder what
400
0:14:15,720 --> 0:14:17,810
are the potential applications for this
401
0:14:17,820 --> 0:14:19,730
kind of software maybe you're far away
402
0:14:19,740 --> 0:14:21,410
from a scene and you want to really look
403
0:14:21,420 --> 0:14:23,090
around the mountains to see if there was
404
0:14:23,100 --> 0:14:24,710
something there maybe you are trying to
405
0:14:24,720 --> 0:14:25,970
look at a crime scene and you want to
406
0:14:25,980 --> 0:14:28,310
see how this scenery is going to look or
407
0:14:28,320 --> 0:14:29,990
how it works under certain previous
408
0:14:30,000 --> 0:14:32,090
conditions this is truly going to change
409
0:14:32,100 --> 0:14:34,129
the game in many different fields and
410
0:14:34,139 --> 0:14:35,750
it's going to be really really effective
411
0:14:35,760 --> 0:14:37,970
I mean I'm someone who spends some time
412
0:14:37,980 --> 0:14:40,250
as well in 3D animating modeling and
413
0:14:40,260 --> 0:14:42,590
rendering so I really do appreciate how
414
0:14:42,600 --> 0:14:44,210
far these tools have come I mean look we
415
0:14:44,220 --> 0:14:45,949
can grab this 2D image right here and
416
0:14:45,959 --> 0:14:47,269
then of course we can click submit
417
0:14:47,279 --> 0:14:48,590
you're about to see that in around a
418
0:14:48,600 --> 0:14:49,850
couple of seconds we're going to get
419
0:14:49,860 --> 0:14:52,069
that 3D mesh reconstruction and on some
420
0:14:52,079 --> 0:14:54,470
images it doesn't work as well but it's
421
0:14:54,480 --> 0:14:56,150
important to note how it does do
422
0:14:56,160 --> 0:14:57,769
different images I think ones where it's
423
0:14:57,779 --> 0:14:59,329
like mountains and stuff like that it
424
0:14:59,339 --> 0:15:01,250
does do really well but nonetheless you
425
0:15:01,260 --> 0:15:02,870
have to understand that once these
426
0:15:02,880 --> 0:15:04,970
images and once these tools get a lot
427
0:15:04,980 --> 0:15:08,150
better the ability to recreate scenes in
428
0:15:08,160 --> 0:15:10,670
3D with photorealistic capabilities are
429
0:15:10,680 --> 0:15:12,410
going to be absolutely insane because
430
0:15:12,420 --> 0:15:14,150
usually essentially if you're trying to
431
0:15:14,160 --> 0:15:16,310
create a scan you need hundreds of image
432
0:15:16,320 --> 0:15:18,290
of a single location but with this one
433
0:15:18,300 --> 0:15:19,910
image we're able to get a decent amount
434
0:15:19,920 --> 0:15:21,590
of information which means that likely
435
0:15:21,600 --> 0:15:23,210
in the future maybe just five or six
436
0:15:23,220 --> 0:15:25,069
images is going to give us a full 3d
437
0:15:25,079 --> 0:15:27,170
reconstruction in high photorealistic
438
0:15:27,180 --> 0:15:28,670
detail and that's something that's not
439
0:15:28,680 --> 0:15:30,230
far away I mean if you take a look at
440
0:15:30,240 --> 0:15:31,730
this image right here this is the final
441
0:15:31,740 --> 0:15:33,230
example I'm going to show you and then
442
0:15:33,240 --> 0:15:35,569
we look at the mesh 3D reconstruction it
443
0:15:35,579 --> 0:15:37,009
just goes to show you what different
444
0:15:37,019 --> 0:15:38,750
parts of the image could be and I mean
445
0:15:38,760 --> 0:15:41,150
that is really really crazy to be able
446
0:15:41,160 --> 0:15:43,189
to look around and have a much more
447
0:15:43,199 --> 0:15:45,590
immersive view of that image I mean just
448
0:15:45,600 --> 0:15:47,689
examining artifacts examining old images
449
0:15:47,699 --> 0:15:49,490
I mean it's just it's just absolutely
450
0:15:49,500 --> 0:15:51,650
crazy I mean maybe I'm over exaggerating
451
0:15:51,660 --> 0:15:53,090
but I do think that this is something
452
0:15:53,100 --> 0:15:55,129
that you definitely should try now this
453
0:15:55,139 --> 0:15:56,990
is the most important part of this video
454
0:15:57,000 --> 0:15:58,550
if there's anything you take away from
455
0:15:58,560 --> 0:16:00,410
this video let it be this short segment
456
0:16:00,420 --> 0:16:01,850
that we are currently talking about okay
457
0:16:01,860 --> 0:16:04,670
this is a research paper Okay produced
458
0:16:04,680 --> 0:16:07,430
by Stanford University and in this
459
0:16:07,440 --> 0:16:09,769
research paper they talk about how chat
460
0:16:09,779 --> 0:16:12,050
gpt's behavior is changing over time
461
0:16:12,060 --> 0:16:14,870
they talk about the fact that chat gbt
462
0:16:14,880 --> 0:16:17,210
has gotten worse over time now at first
463
0:16:17,220 --> 0:16:19,610
I did think this was some kind of rumor
464
0:16:19,620 --> 0:16:21,290
that it was happening but I noticed it
465
0:16:21,300 --> 0:16:23,269
in my own examples you can see here that
466
0:16:23,279 --> 0:16:25,509
in solving math problems it went from a
467
0:16:25,519 --> 0:16:29,509
97.6 accuracy rate to a 2.4 rate for
468
0:16:29,519 --> 0:16:34,430
gpt4 but GPT 3.5 went up from 70.4 to
469
0:16:34,440 --> 0:16:38,389
86.8 in addition you can also see gpt4's
470
0:16:38,399 --> 0:16:40,970
drastic decline on many different tasks
471
0:16:40,980 --> 0:16:43,129
now something also that goes on in this
472
0:16:43,139 --> 0:16:45,410
Twitter thread by Santiago he talks
473
0:16:45,420 --> 0:16:47,629
about how the team used Chain of Thought
474
0:16:47,639 --> 0:16:51,110
to help the model reason is 17077 a
475
0:16:51,120 --> 0:16:53,509
prime number think step by step and if
476
0:16:53,519 --> 0:16:54,590
you've watched the previous videos
477
0:16:54,600 --> 0:16:56,389
you'll understand that Chain of Thought
478
0:16:56,399 --> 0:16:58,009
is a popular technique that
479
0:16:58,019 --> 0:16:59,410
significantly improves answers
480
0:16:59,420 --> 0:17:02,449
unfortunately the latest version of gbt4
481
0:17:02,459 --> 0:17:05,090
did not generate intermediate steps and
482
0:17:05,100 --> 0:17:06,770
instead promptly answered with a simple
483
0:17:06,780 --> 0:17:09,110
note code generation has also gotten
484
0:17:09,120 --> 0:17:11,569
worse the team built a data set with 50
485
0:17:11,579 --> 0:17:13,370
easy problems from Elite code and
486
0:17:13,380 --> 0:17:15,770
measured how many gpt4 answers ran
487
0:17:15,780 --> 0:17:17,449
without any changes the March version
488
0:17:17,459 --> 0:17:20,270
succeeded in 52 of the problems but this
489
0:17:20,280 --> 0:17:23,329
dropped to a pale 10 using the model in
490
0:17:23,339 --> 0:17:25,250
June rumors suggest that they're using
491
0:17:25,260 --> 0:17:27,289
several smaller and specialized gpt4
492
0:17:27,299 --> 0:17:29,510
models that act similarly to a large
493
0:17:29,520 --> 0:17:31,850
model but are less expensive to run when
494
0:17:31,860 --> 0:17:33,650
a user asks a question the system
495
0:17:33,660 --> 0:17:35,750
decides which model to send the query to
496
0:17:35,760 --> 0:17:38,150
cheaper and faster but this new approach
497
0:17:38,160 --> 0:17:39,650
could be the problem behind the
498
0:17:39,660 --> 0:17:41,750
degradation and quality now I also
499
0:17:41,760 --> 0:17:43,549
noticed it as well because a lot of the
500
0:17:43,559 --> 0:17:46,250
time when I was using chat TPT what you
501
0:17:46,260 --> 0:17:48,409
saw in the search bar right here is
502
0:17:48,419 --> 0:17:50,450
giving you exactly which model it was
503
0:17:50,460 --> 0:17:51,650
and you can see right here that the
504
0:17:51,660 --> 0:17:54,230
Vinci 002 was a different model than
505
0:17:54,240 --> 0:17:56,750
chat gbt 3.5 turbo and there were many
506
0:17:56,760 --> 0:17:58,310
different models that we could see when
507
0:17:58,320 --> 0:18:00,529
we went on the chat TPT webpage now it's
508
0:18:00,539 --> 0:18:01,789
not something that was a big story
509
0:18:01,799 --> 0:18:04,010
because when I researched this ages ago
510
0:18:04,020 --> 0:18:05,870
like one or two months ago I noticed
511
0:18:05,880 --> 0:18:07,909
that the post only had three upvotes so
512
0:18:07,919 --> 0:18:09,470
it will be interesting to see if they
513
0:18:09,480 --> 0:18:11,210
managed to update this to reverse the
514
0:18:11,220 --> 0:18:13,010
changes or they start to offer many
515
0:18:13,020 --> 0:18:15,470
different gpt4 models to many different
516
0:18:15,480 --> 0:18:18,169
users in sort of a tier-based way but
517
0:18:18,179 --> 0:18:19,789
this is something that isn't really good
518
0:18:19,799 --> 0:18:21,830
for openai's public relations because
519
0:18:21,840 --> 0:18:23,630
they've done it without telling people
520
0:18:23,640 --> 0:18:25,430
let us know in the comments below if
521
0:18:25,440 --> 0:18:26,450
this is something that you have
522
0:18:26,460 --> 0:18:27,770
personally experienced or something that
523
0:18:27,780 --> 0:18:28,850
you are thinking about and your
524
0:18:28,860 --> 0:18:31,010
suspicions are now confirmed then we had
525
0:18:31,020 --> 0:18:33,169
the United Nations talk about artificial
526
0:18:33,179 --> 0:18:34,789
intelligence and they managed to bring
527
0:18:34,799 --> 0:18:36,590
up some of the key issues that many
528
0:18:36,600 --> 0:18:38,210
people are discussing they talk about
529
0:18:38,220 --> 0:18:40,430
how artificial intelligence could
530
0:18:40,440 --> 0:18:43,310
potentially ruin society and have some
531
0:18:43,320 --> 0:18:45,470
existential impacts that will affect us
532
0:18:45,480 --> 0:18:47,570
all I think it's really important to see
533
0:18:47,580 --> 0:18:49,610
what our world leaders and people and
534
0:18:49,620 --> 0:18:51,289
people who make the very big decisions
535
0:18:51,299 --> 0:18:53,390
or at least influence them have to say
536
0:18:53,400 --> 0:18:56,270
on the very Innovative and moving as
537
0:18:56,280 --> 0:18:58,070
quick as we've ever seen artificial
538
0:18:58,080 --> 0:19:00,289
intelligence space new technology is
539
0:19:00,299 --> 0:19:02,810
moving at warp speeds
540
0:19:02,820 --> 0:19:06,950
and so are the threats that come with it
541
0:19:06,960 --> 0:19:10,070
alarm Bells over the latest form of
542
0:19:10,080 --> 0:19:12,669
artificial intelligence generative AI
543
0:19:12,679 --> 0:19:15,049
are deafening
544
0:19:15,059 --> 0:19:17,810
and they are loudest from the developers
545
0:19:17,820 --> 0:19:19,970
who designed it
546
0:19:19,980 --> 0:19:22,070
the scientists and experts have called
547
0:19:22,080 --> 0:19:23,690
on the World to act
548
0:19:23,700 --> 0:19:26,330
declaring AI an existential threat to
549
0:19:26,340 --> 0:19:27,590
humanity
550
0:19:27,600 --> 0:19:31,070
On a par with the risk of nuclear Wars
551
0:19:31,080 --> 0:19:34,970
we must take those warnings seriously
552
0:19:34,980 --> 0:19:37,370
our proposal Global digital compact new
553
0:19:37,380 --> 0:19:39,650
agenda for peace and Accord on the
554
0:19:39,660 --> 0:19:42,110
global governance of AI will offer
555
0:19:42,120 --> 0:19:44,630
multilateral Solutions based on human
556
0:19:44,640 --> 0:19:46,490
rights
557
0:19:46,500 --> 0:19:49,370
but the Advent of generative AI must not
558
0:19:49,380 --> 0:19:52,909
distract us from the damage digital
559
0:19:52,919 --> 0:19:57,230
technology is already doing to our world
560
0:19:57,240 --> 0:19:59,990
the proliferation of hate and lies in
561
0:20:00,000 --> 0:20:01,490
the digital space
562
0:20:01,500 --> 0:20:05,870
is causing grave Global harm now
563
0:20:05,880 --> 0:20:08,390
it is full in Conflict deaths and
564
0:20:08,400 --> 0:20:10,970
destruction now
565
0:20:10,980 --> 0:20:13,370
it is threatening democracy and human
566
0:20:13,380 --> 0:20:15,830
rights now
567
0:20:15,840 --> 0:20:17,990
it is undermining public health and
568
0:20:18,000 --> 0:20:21,070
climate action now
569
0:20:21,080 --> 0:20:24,950
including some AI innovators have called
570
0:20:24,960 --> 0:20:26,529
for a pause
571
0:20:26,539 --> 0:20:30,169
in the very rapid development of AI
572
0:20:30,179 --> 0:20:32,690
technology before it gets completely out
573
0:20:32,700 --> 0:20:33,830
of control
574
0:20:33,840 --> 0:20:35,590
so that
575
0:20:35,600 --> 0:20:38,570
scientists and policy makers
576
0:20:38,580 --> 0:20:42,950
can assess where this is going and
577
0:20:42,960 --> 0:20:45,909
perhaps suggests some guard rails
578
0:20:45,919 --> 0:20:48,850
possibly including human agency
579
0:20:48,860 --> 0:20:53,870
do you support such a pause and do you
580
0:20:53,880 --> 0:20:55,690
see the
581
0:20:55,700 --> 0:20:58,970
board that you are setting up including
582
0:20:58,980 --> 0:21:04,390
AI scientists as helping in that process
583
0:21:04,400 --> 0:21:10,610
of educating uh during such a pause
584
0:21:10,620 --> 0:21:13,610
I think a pause can be an interesting
585
0:21:13,620 --> 0:21:16,789
idea but I don't think a pause will
586
0:21:16,799 --> 0:21:18,110
solve the problem
587
0:21:18,120 --> 0:21:20,270
because we all know of situations in
588
0:21:20,280 --> 0:21:21,470
which there is a pause and then
589
0:21:21,480 --> 0:21:24,230
everything goes on the same I think that
590
0:21:24,240 --> 0:21:26,330
independently of the the fact that the
591
0:21:26,340 --> 0:21:30,590
pause being a positive possible idea we
592
0:21:30,600 --> 0:21:32,630
need to make sure that we move forward
593
0:21:32,640 --> 0:21:34,130
and we move forward as quickly as
594
0:21:34,140 --> 0:21:35,990
possible in all the other mechanisms
595
0:21:36,000 --> 0:21:38,090
that we have been discussing great thank
596
0:21:38,100 --> 0:21:39,649
you very much sir thank you Melissa
597
0:21:39,659 --> 0:21:41,149
thank you
598
0:21:41,159 --> 0:21:46,400
and we'll see you back in uh around 12.
599
0:21:48,780 --> 0:21:52,190
Kristen Pam James Bays from Al Jazeera
600
0:21:52,200 --> 0:21:53,390
um Secretary General you've been
601
0:21:53,400 --> 0:21:55,850
speaking about AI for quite a long time
602
0:21:55,860 --> 0:21:57,590
and I'm sure been thinking about it as
603
0:21:57,600 --> 0:21:58,970
well in your opening statement you
604
0:21:58,980 --> 0:22:01,430
quoted scientists and experts saying
605
0:22:01,440 --> 0:22:03,230
that it could be a risk as high as
606
0:22:03,240 --> 0:22:05,270
nuclear war
607
0:22:05,280 --> 0:22:07,010
um some are also though saying that it
608
0:22:07,020 --> 0:22:09,289
could cure cancer so I just like your
609
0:22:09,299 --> 0:22:11,510
assessment what do you think are the
610
0:22:11,520 --> 0:22:15,230
challenges and the opportunities of AI
611
0:22:15,240 --> 0:22:18,049
I think the opportunities are immense if
612
0:22:18,059 --> 0:22:20,690
you look at Health if you look at the
613
0:22:20,700 --> 0:22:23,649
environment if you look at education
614
0:22:23,659 --> 0:22:29,330
what AI can produce can be an extremely
615
0:22:29,340 --> 0:22:31,850
important factor to make the sustainable
616
0:22:31,860 --> 0:22:36,169
development goals a reality so AI has an
617
0:22:36,179 --> 0:22:37,730
enormous potential
618
0:22:37,740 --> 0:22:42,110
but it is clear that AI
619
0:22:42,120 --> 0:22:45,529
as a serious problem which is the
620
0:22:45,539 --> 0:22:48,289
possible removal of human agency
621
0:22:48,299 --> 0:22:51,110
and that for me is the central question
622
0:22:51,120 --> 0:22:54,529
it is absolutely essential that human
623
0:22:54,539 --> 0:22:57,710
agency remains present in everything
624
0:22:57,720 --> 0:23:00,830
that is built with artificial
625
0:23:00,840 --> 0:23:02,570
intelligence
626
0:23:02,580 --> 0:23:04,789
but as I said in the beginning
627
0:23:04,799 --> 0:23:07,490
the fact that we look at catastrophic
628
0:23:07,500 --> 0:23:10,310
consequences might come from artificial
629
0:23:10,320 --> 0:23:12,830
intelligence in the future should not
630
0:23:12,840 --> 0:23:15,529
distract us from what's happening today
631
0:23:15,539 --> 0:23:18,350
in the digital world that is
632
0:23:18,360 --> 0:23:20,870
contributing to people being killed to
633
0:23:20,880 --> 0:23:23,450
Human Rights be violated to our privacy
634
0:23:23,460 --> 0:23:24,430
being
635
0:23:24,440 --> 0:23:27,130
completely
636
0:23:27,140 --> 0:23:32,330
destroyed and for the data that is
637
0:23:32,340 --> 0:23:35,870
produced by us to fully Escape our
638
0:23:35,880 --> 0:23:37,909
control so many things are happening
639
0:23:37,919 --> 0:23:41,450
today that we need to deal with
640
0:23:41,460 --> 0:23:44,149
but at the same time we need to do
641
0:23:44,159 --> 0:23:46,549
everything possible to make sure that
642
0:23:46,559 --> 0:23:49,270
the future evolution of AI doesn't go
643
0:23:49,280 --> 0:23:52,970
into the logic that completely abolishes
644
0:23:52,980 --> 0:23:56,270
human agency and creates a monster that
645
0:23:56,280 --> 0:23:58,820
nobody will be able
44989
Can't find what you're looking for?
Get subtitles in any language from opensubtitles.com, and translate them here.
Afrikaans
Albanian
Amharic
Armenian
Azerbaijani
Basque
Belarusian
Bengali
Bosnian
Bulgarian
Catalan
Cebuano
Chichewa
Chinese (Simplified)
Chinese (Traditional)
Corsican
Croatian
Czech
Danish
Dutch
English
Esperanto
Estonian
Finnish
French
Frisian
Galician
Georgian
German
Greek
Gujarati
Haitian Creole
Hawaiian
Hungarian
Icelandic
Indonesian
Irish
Italian
Japanese
Kazakh
Khmer
Korean
Kurdish (Kurmanji)
Kyrgyz
Lao
Latvian
Lithuanian
Luxembourgish
Macedonian
Malagasy
Malay
Maltese
Maori
Mongolian
Myanmar (Burmese)
Nepali
Norwegian
Pashto
Polish
Portuguese
Romanian
Russian
Samoan
Serbian
Sesotho
Shona
Sinhala
Slovak
Slovenian
Somali
Spanish
Swahili
Swedish
Tajik
Thai
Turkish
Ukrainian
Urdu
Uzbek
Vietnamese
