Skip to content

Instantly share code, notes, and snippets.

Show Gist options
  • Save JimLiu/0564678cf555794fda77ca0bb312eb98 to your computer and use it in GitHub Desktop.
Save JimLiu/0564678cf555794fda77ca0bb312eb98 to your computer and use it in GitHub Desktop.
How AI Could Empower Any Business | Andrew Ng | TED
[Script Info]
Title: How AI Could Empower Any Business | Andrew Ng | TED
ScriptType: v4.00+
WrapStyle: 0
Collisions: Reverse
PlayResX: 384
PlayResY: 288
Timer: 100.0000
ScaledBorderAndShadow: no
Last Style Storage: Default
Video Aspect Ratio: 0
Video Zoom: 6
Video Position: 0
[V4+ Styles]
Format: Name,Fontname,Fontsize,PrimaryColour,SecondaryColour,OutlineColour,BackColour,Bold,Italic,Underline,StrikeOut,ScaleX,ScaleY,Spacing,Angle,BorderStyle,Outline,Shadow,Alignment,MarginL,MarginR,MarginV,Encoding
Style: Default,HONOR Sans CN,20,&H0080FFFF,&H000000FF,&H00000000,&H00000000,-1,0,0,0,100,100,1,0,1,2,0,2,1,1,6,1
Style: Secondary,Helvetica,12,&H00FFFFFF,&H000000FF,&H00000000,&H00000000,-1,0,0,0,100,100,0,0,1,2,0,2,1,1,6,1
[Events]
Format: Layer, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text
Dialogue: 9,0:00:05.0,0:00:10.0,Secondary,,0,0,0,,{\an7\fs12\pos(9,11)\fad(300,1000)}{\1c&H00FFFFFF&\2c&H0000FF&}翻译:{\1c&H80FFFF&\2c&H0000FF&}宝玉
Dialogue: 9,0:00:00.0,0:00:05.0,Default,,0,0,0,,{\fs20\an8}{\fad(200,200)\c&HFFFFFF&}How AI Could Empower Any Business | Andrew Ng | TED
Dialogue: 1,0:00:04.60,0:00:10.88,Secondary,,0,0,0,,When I think about the rise of AI, I'm reminded by the rise of literacy.
Dialogue: 1,0:00:10.88,0:00:12.19,Secondary,,0,0,0,,A few hundred years ago,
Dialogue: 1,0:00:12.37,0:00:13.67,Secondary,,0,0,0,,many people in society thought
Dialogue: 1,0:00:13.89,0:00:17.65,Secondary,,0,0,0,,that maybe not everyone needed to be able to read and write.
Dialogue: 1,0:00:17.65,0:00:20.35,Secondary,,0,0,0,,Back then, many people were attending fields or herding sheep,
Dialogue: 1,0:00:20.49,0:00:22.43,Secondary,,0,0,0,,so maybe there was less need for written communication.
Dialogue: 1,0:00:23.3,0:00:26.44,Secondary,,0,0,0,,And all that was needed was for the high priests and priestesses and monks
Dialogue: 1,0:00:26.54,0:00:28.44,Secondary,,0,0,0,,to be able to read the holy book,
Dialogue: 1,0:00:28.54,0:00:32.36,Secondary,,0,0,0,,and the rest of us could just go to the temple or church or the holy building
Dialogue: 1,0:00:32.44,0:00:36.6,Secondary,,0,0,0,,and sit and listen to the high priests and priestesses read to us.
Dialogue: 1,0:00:36.6,0:00:39.41,Secondary,,0,0,0,,Fortunately, it was since figured out that we can build a much richer society
Dialogue: 1,0:00:39.48,0:00:41.23,Secondary,,0,0,0,,if lots of people can read and write.
Dialogue: 1,0:00:42.23,0:00:46.85,Secondary,,0,0,0,,Today, AI is in the hands of the high priests and priestesses.
Dialogue: 1,0:00:46.85,0:00:48.69,Secondary,,0,0,0,,These are the highly skilled AI engineers,
Dialogue: 1,0:00:48.85,0:00:51.81,Secondary,,0,0,0,,many of whom work in the big tech companies.
Dialogue: 1,0:00:51.81,0:00:56.11,Secondary,,0,0,0,,And most people have access only to the AI that they built for them.
Dialogue: 1,0:00:56.11,0:00:58.53,Secondary,,0,0,0,,I think that we can build a much richer society
Dialogue: 1,0:00:58.59,0:01:02.58,Secondary,,0,0,0,,if we can enable everyone to help to write the future.
Dialogue: 1,0:01:04.12,0:01:08.55,Secondary,,0,0,0,,But why is AI largely concentrated in the big tech companies?
Dialogue: 1,0:01:08.55,0:01:11.39,Secondary,,0,0,0,,Because many of these AI projects have been expensive to build.
Dialogue: 1,0:01:11.39,0:01:14.50,Secondary,,0,0,0,,They may require dozens of highly skilled engineers,
Dialogue: 1,0:01:14.58,0:01:18.22,Secondary,,0,0,0,,and it may cost millions or tens of millions of dollars to build an AI system.
Dialogue: 1,0:01:19.30,0:01:20.72,Secondary,,0,0,0,,And the large tech companies,
Dialogue: 1,0:01:21.4,0:01:24.64,Secondary,,0,0,0,,particularly the ones with hundreds of millions or even billions of users,
Dialogue: 1,0:01:24.76,0:01:29.1,Secondary,,0,0,0,,have been better than anyone else at making these investments pay off,
Dialogue: 1,0:01:29.55,0:01:33.47,Secondary,,0,0,0,,because for them, a one-size-fits-all AI system,
Dialogue: 1,0:01:33.67,0:01:35.51,Secondary,,0,0,0,,such as one that improves web search
Dialogue: 1,0:01:35.59,0:01:38.49,Secondary,,0,0,0,,or that recommends better products for online shopping,
Dialogue: 1,0:01:38.65,0:01:41.61,Secondary,,0,0,0,,can be applied to this very large number of users
Dialogue: 1,0:01:41.71,0:01:43.49,Secondary,,0,0,0,,to generate a massive amount of revenue.
Dialogue: 1,0:01:44.64,0:01:48.12,Secondary,,0,0,0,,But this recipe for AI does not work
Dialogue: 1,0:01:48.22,0:01:52.84,Secondary,,0,0,0,,once you go outside the tech and internet sectors to other places where,
Dialogue: 1,0:01:52.94,0:01:55.38,Secondary,,0,0,0,,for the most part, there are hardly any projects
Dialogue: 1,0:01:55.50,0:02:01.14,Secondary,,0,0,0,,that apply to 100 million people or that generate comparable economics.
Dialogue: 1,0:02:01.14,0:02:02.41,Secondary,,0,0,0,,Let me illustrate an example.
Dialogue: 1,0:02:03.92,0:02:05.72,Secondary,,0,0,0,,Many weekends,
Dialogue: 1,0:02:05.82,0:02:09.32,Secondary,,0,0,0,,I drive a few minutes from my house to a local pizza store
Dialogue: 1,0:02:09.44,0:02:14.63,Secondary,,0,0,0,,to buy a slice of Hawaiian pizza from the gentleman that owns his pizza store.
Dialogue: 1,0:02:14.63,0:02:15.87,Secondary,,0,0,0,,And his pizza's great,
Dialogue: 1,0:02:16.5,0:02:19.39,Secondary,,0,0,0,,but he always has a lot of cold pizza sitting around,
Dialogue: 1,0:02:19.53,0:02:22.31,Secondary,,0,0,0,,and every weekend, some different flavor of pizza is out of stock.
Dialogue: 1,0:02:23.29,0:02:25.99,Secondary,,0,0,0,,But when I watch him operate his store,
Dialogue: 1,0:02:26.7,0:02:29.55,Secondary,,0,0,0,,I get excited, because by selling pizza,
Dialogue: 1,0:02:29.67,0:02:32.13,Secondary,,0,0,0,,he is generating data,
Dialogue: 1,0:02:32.23,0:02:38.16,Secondary,,0,0,0,,and this is data that he can take advantage of if he had access to AI.
Dialogue: 1,0:02:38.16,0:02:43.36,Secondary,,0,0,0,,AI systems are good at spotting patterns when given access to the right data,
Dialogue: 1,0:02:43.44,0:02:45.22,Secondary,,0,0,0,,and perhaps an AI system could spot
Dialogue: 1,0:02:45.30,0:02:49.10,Secondary,,0,0,0,,if Mediterranean pizzas sell really well on a Friday night.
Dialogue: 1,0:02:49.10,0:02:52.43,Secondary,,0,0,0,,Maybe it could suggest to him to make more of it on a Friday afternoon.
Dialogue: 1,0:02:53.46,0:02:54.26,Secondary,,0,0,0,,Now, you might say to me,
Dialogue: 1,0:02:54.40,0:02:58.26,Secondary,,0,0,0,,hey, Andrew, this is a small pizza store, what's the big deal?
Dialogue: 1,0:02:58.26,0:03:01.59,Secondary,,0,0,0,,And I say to the gentleman that owns this pizza store,
Dialogue: 1,0:03:01.85,0:03:03.65,Secondary,,0,0,0,,something that could help him improve his revenues
Dialogue: 1,0:03:03.75,0:03:05.23,Secondary,,0,0,0,,by a few thousand dollars a year,
Dialogue: 1,0:03:05.37,0:03:08.95,Secondary,,0,0,0,,that would be a huge deal to him.
Dialogue: 1,0:03:08.95,0:03:14.56,Secondary,,0,0,0,,I know that there is a lot of hype about AI's need for massive data sets.
Dialogue: 1,0:03:14.56,0:03:15.96,Secondary,,0,0,0,,And having more data does help.
Dialogue: 1,0:03:17.23,0:03:19.47,Secondary,,0,0,0,,But contrary to the hype,
Dialogue: 1,0:03:19.81,0:03:23.32,Secondary,,0,0,0,,AI can often work just fine even on modest amounts of data,
Dialogue: 1,0:03:23.46,0:03:26.84,Secondary,,0,0,0,,such as the data generated by a single pizza store.
Dialogue: 1,0:03:26.84,0:03:30.92,Secondary,,0,0,0,,So the real problem is not that there isn't enough data from the pizza store.
Dialogue: 1,0:03:30.92,0:03:34.86,Secondary,,0,0,0,,The real problem is that the small pizza store could never serve enough customers
Dialogue: 1,0:03:34.96,0:03:37.92,Secondary,,0,0,0,,to justify the cost of hiring an AI team.
Dialogue: 1,0:03:39.55,0:03:42.1,Secondary,,0,0,0,,I know that in the United States,
Dialogue: 1,0:03:42.13,0:03:44.43,Secondary,,0,0,0,,there are about half a million independent restaurants,
Dialogue: 1,0:03:44.51,0:03:48.57,Secondary,,0,0,0,,and collectively, these restaurants do serve tens of millions of customers,
Dialogue: 1,0:03:48.73,0:03:50.21,Secondary,,0,0,0,,but every restaurant is different,
Dialogue: 1,0:03:50.37,0:03:53.81,Secondary,,0,0,0,,with a different menu, different customers, different ways of recording sales,
Dialogue: 1,0:03:53.95,0:03:56.99,Secondary,,0,0,0,,that no one-size-fits-all AI will work for all of them.
Dialogue: 1,0:03:58.66,0:04:02.12,Secondary,,0,0,0,,What would it be like if we could enable small businesses,
Dialogue: 1,0:04:02.20,0:04:05.44,Secondary,,0,0,0,,and especially local businesses, to use AI?
Dialogue: 1,0:04:05.44,0:04:08.22,Secondary,,0,0,0,,Let's take a look at what it might look like at a company
Dialogue: 1,0:04:08.37,0:04:10.91,Secondary,,0,0,0,,that makes and sells T-shirts.
Dialogue: 1,0:04:10.91,0:04:14.39,Secondary,,0,0,0,,I would love if an accountant working for the T-shirt company
Dialogue: 1,0:04:14.53,0:04:16.79,Secondary,,0,0,0,,can use AI for demand forecasting.
Dialogue: 1,0:04:16.79,0:04:19.57,Secondary,,0,0,0,,Say, figure out what funny memes to print on T-shirts
Dialogue: 1,0:04:19.69,0:04:22.27,Secondary,,0,0,0,,that would drive sales by looking at what's trending on social media.
Dialogue: 1,0:04:23.37,0:04:25.43,Secondary,,0,0,0,,Or for product placement,
Dialogue: 1,0:04:25.59,0:04:29.60,Secondary,,0,0,0,,why can't a front-of-store manager take pictures of what the store looks like
Dialogue: 1,0:04:29.68,0:04:31.90,Secondary,,0,0,0,,and show it to an AI and have an AI recommend
Dialogue: 1,0:04:32.6,0:04:34.90,Secondary,,0,0,0,,where to place products to improve sales?
Dialogue: 1,0:04:34.90,0:04:37.81,Secondary,,0,0,0,,Supply chain, can an AI recommend to a buyer
Dialogue: 1,0:04:38.3,0:04:42.19,Secondary,,0,0,0,,whether or not they should pay 20 dollars per yard for a piece of fabric now
Dialogue: 1,0:04:42.25,0:04:43.37,Secondary,,0,0,0,,or if they should keep looking
Dialogue: 1,0:04:43.61,0:04:45.21,Secondary,,0,0,0,,because they might be able to find it cheaper elsewhere?
Dialogue: 1,0:04:46.34,0:04:47.86,Secondary,,0,0,0,,Or quality control.
Dialogue: 1,0:04:47.86,0:04:50.68,Secondary,,0,0,0,,A quality inspector should be able to use AI
Dialogue: 1,0:04:50.86,0:04:55.40,Secondary,,0,0,0,,to automatically scan pictures of the fabric being used to make T-shirts
Dialogue: 1,0:04:55.50,0:04:59.35,Secondary,,0,0,0,,to check if there are any tears or discolorations in the cloth.
Dialogue: 1,0:04:59.35,0:05:02.71,Secondary,,0,0,0,,Today, large tech companies routinely use AI
Dialogue: 1,0:05:02.81,0:05:05.45,Secondary,,0,0,0,,to solve problems like these and to great effect.
Dialogue: 1,0:05:06.79,0:05:11.79,Secondary,,0,0,0,,But a typical T-shirt company or a typical auto mechanic
Dialogue: 1,0:05:11.87,0:05:16.15,Secondary,,0,0,0,,or retailer or school or local farm will be using AI
Dialogue: 1,0:05:16.27,0:05:20.2,Secondary,,0,0,0,,for exactly zero of these applications today.
Dialogue: 1,0:05:20.2,0:05:24.24,Secondary,,0,0,0,,Every T-shirt maker is sufficiently different from every other T-shirt maker
Dialogue: 1,0:05:24.38,0:05:27.22,Secondary,,0,0,0,,that there is no one-size-fits-all AI that will work for all of them.
Dialogue: 1,0:05:28.34,0:05:33.34,Secondary,,0,0,0,,And in fact, once you go outside the internet and tech sectors,
Dialogue: 1,0:05:33.42,0:05:35.85,Secondary,,0,0,0,,in other industries, even large companies,
Dialogue: 1,0:05:36.3,0:05:39.81,Secondary,,0,0,0,,such as the pharmaceutical companies, the carmakers, the hospitals,
Dialogue: 1,0:05:40.13,0:05:42.65,Secondary,,0,0,0,,also struggle with this.
Dialogue: 1,0:05:42.65,0:05:45.31,Secondary,,0,0,0,,This is the long-tail problem of AI.
Dialogue: 1,0:05:46.57,0:05:51.14,Secondary,,0,0,0,,If you were to take all current and potential AI projects
Dialogue: 1,0:05:51.22,0:05:55.28,Secondary,,0,0,0,,and sort them in decreasing order of value and plot them,
Dialogue: 1,0:05:55.42,0:05:57.22,Secondary,,0,0,0,,you'd get a graph that looks like this.
Dialogue: 1,0:05:57.22,0:05:59.84,Secondary,,0,0,0,,Maybe the single most valuable AI system is something
Dialogue: 1,0:05:59.100,0:06:02.64,Secondary,,0,0,0,,that decides what ads to show people on the internet.
Dialogue: 1,0:06:02.64,0:06:04.96,Secondary,,0,0,0,,Maybe the second most valuable is a web search engine.
Dialogue: 1,0:06:04.96,0:06:06.28,Secondary,,0,0,0,,Maybe the third most valuable
Dialogue: 1,0:06:06.36,0:06:08.46,Secondary,,0,0,0,,is an online shopping product recommendation system.
Dialogue: 1,0:06:09.57,0:06:12.35,Secondary,,0,0,0,,But when you go to the right of this curve,
Dialogue: 1,0:06:12.55,0:06:15.67,Secondary,,0,0,0,,you then get projects like T-shirt product placements
Dialogue: 1,0:06:15.75,0:06:20.25,Secondary,,0,0,0,,or T-shirt demand forecasting or Pizzeria demand forecasting.
Dialogue: 1,0:06:20.25,0:06:25.5,Secondary,,0,0,0,,And each of these is a unique project that needs to be custom-built.
Dialogue: 1,0:06:25.5,0:06:26.83,Secondary,,0,0,0,,Even T-shirt demand forecasting,
Dialogue: 1,0:06:26.89,0:06:30.9,Secondary,,0,0,0,,if it depends on trending memes on social media,
Dialogue: 1,0:06:30.17,0:06:34.43,Secondary,,0,0,0,,is a very different project than Pizzeria demand forecasting,
Dialogue: 1,0:06:34.49,0:06:36.39,Secondary,,0,0,0,,if that depends on the Pizzeria sales data.
Dialogue: 1,0:06:37.44,0:06:42.9,Secondary,,0,0,0,,So today, there are millions of projects sitting on the tail of this distribution
Dialogue: 1,0:06:42.21,0:06:46.65,Secondary,,0,0,0,,that no one is working on, but whose aggregate value is massive.
Dialogue: 1,0:06:46.65,0:06:49.71,Secondary,,0,0,0,,So how can we enable small businesses and individuals
Dialogue: 1,0:06:49.79,0:06:51.83,Secondary,,0,0,0,,to build AI systems that matter to them?
Dialogue: 1,0:06:53.22,0:06:54.96,Secondary,,0,0,0,,For most of the last few decades,
Dialogue: 1,0:06:55.4,0:06:57.34,Secondary,,0,0,0,,if you want to build an AI system,
Dialogue: 1,0:06:57.48,0:06:58.50,Secondary,,0,0,0,,this is what you have to do.
Dialogue: 1,0:06:58.50,0:07:01.21,Secondary,,0,0,0,,You have to write pages and pages of code.
Dialogue: 1,0:07:01.21,0:07:03.99,Secondary,,0,0,0,,And while I would love for everyone to learn to code,
Dialogue: 1,0:07:04.9,0:07:04.77,Secondary,,0,0,0,,and in fact,
Dialogue: 1,0:07:05.7,0:07:07.11,Secondary,,0,0,0,,online education and also offline education
Dialogue: 1,0:07:07.19,0:07:10.1,Secondary,,0,0,0,,are helping more people than ever learn to code,
Dialogue: 1,0:07:10.62,0:07:12.86,Secondary,,0,0,0,,unfortunately, not everyone has the time to do this.
Dialogue: 1,0:07:13.87,0:07:17.63,Secondary,,0,0,0,,But there is an emerging new way to build AI systems
Dialogue: 1,0:07:17.75,0:07:20.13,Secondary,,0,0,0,,that will let more people participate.
Dialogue: 1,0:07:20.13,0:07:25.40,Secondary,,0,0,0,,Just as pen and paper, which are a vastly superior technology to stone,
Dialogue: 1,0:07:25.84,0:07:29.78,Secondary,,0,0,0,,tablet and chisel, were instrumental to widespread literacy,
Dialogue: 1,0:07:29.92,0:07:32.82,Secondary,,0,0,0,,there are emerging new AI development platforms
Dialogue: 1,0:07:32.96,0:07:36.8,Secondary,,0,0,0,,that shift the focus from asking you to write lots of code
Dialogue: 1,0:07:36.24,0:07:39.30,Secondary,,0,0,0,,to asking you to focus on providing data,
Dialogue: 1,0:07:39.36,0:07:41.98,Secondary,,0,0,0,,and this turns out to be much easier for a lot of people to do.
Dialogue: 1,0:07:43.27,0:07:47.88,Secondary,,0,0,0,,Today, there are multiple companies working on platforms like these.
Dialogue: 1,0:07:47.88,0:07:49.66,Secondary,,0,0,0,,Let me illustrate a few of the concepts
Dialogue: 1,0:07:49.86,0:07:52.12,Secondary,,0,0,0,,using one that my team has been building.
Dialogue: 1,0:07:52.12,0:07:58.52,Secondary,,0,0,0,,Take the example of an inspector wanting AI to help detect defects in fabric.
Dialogue: 1,0:07:58.52,0:08:00.60,Secondary,,0,0,0,,An inspector can take pictures of the fabric
Dialogue: 1,0:08:00.68,0:08:02.68,Secondary,,0,0,0,,and upload it to a platform like this.
Dialogue: 1,0:08:03.56,0:08:05.74,Secondary,,0,0,0,,And they can go in to show the AI
Dialogue: 1,0:08:05.90,0:08:09.43,Secondary,,0,0,0,,what tears in the fabric look like by drawing rectangles.
Dialogue: 1,0:08:09.43,0:08:11.71,Secondary,,0,0,0,,And they can also go in to show the AI
Dialogue: 1,0:08:11.87,0:08:16.25,Secondary,,0,0,0,,what discolorations in the fabric look like by drawing rectangles.
Dialogue: 1,0:08:16.25,0:08:17.55,Secondary,,0,0,0,,So these pictures,
Dialogue: 1,0:08:18.3,0:08:21.42,Secondary,,0,0,0,,together with the green and pink rectangles that the inspector has drawn,
Dialogue: 1,0:08:21.52,0:08:24.94,Secondary,,0,0,0,,are data created by the inspector to explain to the AI
Dialogue: 1,0:08:25.8,0:08:27.12,Secondary,,0,0,0,,how to find tears and discolorations.
Dialogue: 1,0:08:28.59,0:08:30.45,Secondary,,0,0,0,,After the AI examines this data,
Dialogue: 1,0:08:30.55,0:08:33.2,Secondary,,0,0,0,,we may find that it has seen enough pictures of tears,
Dialogue: 1,0:08:33.12,0:08:35.52,Secondary,,0,0,0,,but not yet enough pictures of discolorations.
Dialogue: 1,0:08:35.52,0:08:39.92,Secondary,,0,0,0,,This is akin to if a junior inspector had learned to reliably spot tears,
Dialogue: 1,0:08:40.6,0:08:43.70,Secondary,,0,0,0,,but still needs to further hone their judgment about discolorations.
Dialogue: 1,0:08:43.70,0:08:45.15,Secondary,,0,0,0,,So the inspector can go back
Dialogue: 1,0:08:45.23,0:08:48.45,Secondary,,0,0,0,,and take more pictures of discolorations to show to the AI
Dialogue: 1,0:08:48.57,0:08:50.73,Secondary,,0,0,0,,to help it deepen its understanding.
Dialogue: 1,0:08:50.73,0:08:53.33,Secondary,,0,0,0,,By adjusting the data you give to the AI,
Dialogue: 1,0:08:53.43,0:08:55.15,Secondary,,0,0,0,,you can help the AI get smarter.
Dialogue: 1,0:08:56.48,0:09:00.90,Secondary,,0,0,0,,So an inspector using an accessible platform like this can,
Dialogue: 1,0:09:00.98,0:09:03.48,Secondary,,0,0,0,,in a few hours to a few days,
Dialogue: 1,0:09:03.58,0:09:07.16,Secondary,,0,0,0,,and with purchasing a suitable camera setup,
Dialogue: 1,0:09:07.26,0:09:11.83,Secondary,,0,0,0,,be able to build a custom AI system to detect defects,
Dialogue: 1,0:09:12.15,0:09:14.45,Secondary,,0,0,0,,tears and discrepancies in all the fabric being used
Dialogue: 1,0:09:14.51,0:09:16.71,Secondary,,0,0,0,,to make T-shirts throughout the factory.
Dialogue: 1,0:09:16.71,0:09:19.75,Secondary,,0,0,0,,And once again, you may say, hey, Andrew,
Dialogue: 1,0:09:20.82,0:09:22.0,Secondary,,0,0,0,,This is one factory.
Dialogue: 1,0:09:22.0,0:09:24.12,Secondary,,0,0,0,,Why is this a big deal?
Dialogue: 1,0:09:24.12,0:09:24.86,Secondary,,0,0,0,,And I say to you,
Dialogue: 1,0:09:24.92,0:09:28.36,Secondary,,0,0,0,,is this a big deal to that inspector whose life this makes easier?
Dialogue: 1,0:09:28.36,0:09:32.27,Secondary,,0,0,0,,And equally, this type of technology can empower a baker
Dialogue: 1,0:09:32.35,0:09:35.47,Secondary,,0,0,0,,to use AI to check for the quality of the cakes they're making,
Dialogue: 1,0:09:35.55,0:09:39.51,Secondary,,0,0,0,,or an organic farmer to check the quality of the vegetables,
Dialogue: 1,0:09:39.61,0:09:42.85,Secondary,,0,0,0,,or a furniture maker to check the quality of the wood they're using.
Dialogue: 1,0:09:44.90,0:09:47.90,Secondary,,0,0,0,,Platforms like these will probably still need a few more years
Dialogue: 1,0:09:48.19,0:09:51.87,Secondary,,0,0,0,,before they're easy enough to use for every pizzeria owner.
Dialogue: 1,0:09:51.87,0:09:54.7,Secondary,,0,0,0,,But many of these platforms are coming along,
Dialogue: 1,0:09:54.15,0:09:56.76,Secondary,,0,0,0,,and some of them are getting to be quite usable to someone
Dialogue: 1,0:09:56.86,0:10:00.4,Secondary,,0,0,0,,that is tech-savvy today with just a bit of training.
Dialogue: 1,0:10:00.74,0:10:04.96,Secondary,,0,0,0,,But what this means is that rather than relying on the high priests and priestesses
Dialogue: 1,0:10:05.6,0:10:07.78,Secondary,,0,0,0,,to write AI systems for everyone else,
Dialogue: 1,0:10:07.88,0:10:10.26,Secondary,,0,0,0,,we can start to empower every accountant,
Dialogue: 1,0:10:10.40,0:10:12.73,Secondary,,0,0,0,,every store manager, every buyer,
Dialogue: 1,0:10:12.83,0:10:16.19,Secondary,,0,0,0,,and every quality inspector to build their own AI systems.
Dialogue: 1,0:10:17.55,0:10:22.49,Secondary,,0,0,0,,I hope that the pizzeria owner and many other small business owners like him
Dialogue: 1,0:10:22.63,0:10:25.1,Secondary,,0,0,0,,will also take advantage of this technology,
Dialogue: 1,0:10:25.47,0:10:28.55,Secondary,,0,0,0,,because AI is creating tremendous wealth
Dialogue: 1,0:10:28.63,0:10:30.94,Secondary,,0,0,0,,and will continue to create tremendous wealth.
Dialogue: 1,0:10:30.94,0:10:33.80,Secondary,,0,0,0,,And it's only by democratizing access to AI
Dialogue: 1,0:10:33.90,0:10:39.86,Secondary,,0,0,0,,that we can ensure that this wealth is spread far and wide across society.
Dialogue: 1,0:10:39.86,0:10:42.47,Secondary,,0,0,0,,Hundreds of years ago, I think hardly anyone
Dialogue: 1,0:10:43.56,0:10:47.68,Secondary,,0,0,0,,understood the impact that widespread literacy will have.
Dialogue: 1,0:10:47.68,0:10:51.26,Secondary,,0,0,0,,Today, I think hardly anyone understands the impact
Dialogue: 1,0:10:51.42,0:10:55.11,Secondary,,0,0,0,,that democratizing access to AI will have.
Dialogue: 1,0:10:55.11,0:10:58.95,Secondary,,0,0,0,,Building AI systems has been out of reach for most people,
Dialogue: 1,0:10:59.7,0:11:01.43,Secondary,,0,0,0,,but that does not have to be the case.
Dialogue: 1,0:11:01.43,0:11:03.33,Secondary,,0,0,0,,In the coming era for AI,
Dialogue: 1,0:11:03.49,0:11:07.11,Secondary,,0,0,0,,we'll empower everyone to build AI systems for themselves,
Dialogue: 1,0:11:07.17,0:11:10.15,Secondary,,0,0,0,,and I think that will be an incredibly exciting future.
Dialogue: 1,0:11:10.15,0:11:10.83,Secondary,,0,0,0,,Thank you very much.
Dialogue: 1,0:00:04.60,0:00:10.88,Default,,0,0,0,,当我想到 AI (人工智能)的崛起之时, 我联想了读写能力的崛起。
Dialogue: 1,0:00:10.88,0:00:13.67,Default,,0,0,0,,几百年前,社会上的许多人认为,也许不是每个人都需要会读写。
Dialogue: 1,0:00:13.89,0:00:22.43,Default,,0,0,0,,那时候,许多人在田地里工作或放羊,所以对书面交流的需求较少。
Dialogue: 1,0:00:23.3,0:00:28.44,Default,,0,0,0,,只有高级祭司、僧侣能够读得懂圣书,
Dialogue: 1,0:00:28.54,0:00:32.36,Default,,0,0,0,,而其他人只需去寺庙、教堂或圣殿,
Dialogue: 1,0:00:32.44,0:00:36.6,Default,,0,0,0,,坐下来听高级祭司读给我们听。
Dialogue: 1,0:00:36.6,0:00:41.23,Default,,0,0,0,,幸运的是,我们后来发现,如果许多人都会读\N写,我们可以建立一个更加富裕的社会。
Dialogue: 1,0:00:42.23,0:00:46.85,Default,,0,0,0,,如今,人工智能掌握在高级祭司和僧侣的手中。
Dialogue: 1,0:00:46.85,0:00:48.69,Default,,0,0,0,,这些高技能的人工智能工程师,
Dialogue: 1,0:00:48.85,0:00:51.81,Default,,0,0,0,,其中很多就职于科技巨头公司。
Dialogue: 1,0:00:51.81,0:00:56.11,Default,,0,0,0,,而大多数人只能接触到这些工程师构建的人工智能。
Dialogue: 1,0:00:56.11,0:01:02.58,Default,,0,0,0,,我认为,如果我们能让每个人参与谱写未\N来,我们就能创造一个更富裕的社会。
Dialogue: 1,0:01:04.12,0:01:08.55,Default,,0,0,0,,但是为什么人工智能主要集中在科技巨头手中呢?
Dialogue: 1,0:01:08.55,0:01:11.39,Default,,0,0,0,,因为开发这些人工智能项目的成本很高。
Dialogue: 1,0:01:11.39,0:01:14.50,Default,,0,0,0,,这些项目需要一大群技术高超的工程师,
Dialogue: 1,0:01:14.58,0:01:18.22,Default,,0,0,0,,要开发一个人工智能系统可能要数百万甚至数千万美元。
Dialogue: 1,0:01:19.30,0:01:24.64,Default,,0,0,0,,大型科技公司,尤其是拥有数亿甚至数十亿用户的公司,
Dialogue: 1,0:01:24.76,0:01:29.1,Default,,0,0,0,,最擅长收回这些投资,
Dialogue: 1,0:01:29.55,0:01:33.47,Default,,0,0,0,,因为对于他们来说,一个适用于所有人的人工智能系统,
Dialogue: 1,0:01:33.67,0:01:38.49,Default,,0,0,0,,比如改进网络搜索或为在线购物推荐更好的产品,
Dialogue: 1,0:01:38.65,0:01:43.49,Default,,0,0,0,,可以应用于大量用户,从而产生巨额收益。
Dialogue: 1,0:01:44.64,0:01:52.84,Default,,0,0,0,,但是一旦你走出科技互联网行业,进入其他领\N域, 这个人工智能的模式可能就不奏效了,
Dialogue: 1,0:01:52.94,0:02:01.14,Default,,0,0,0,,因为在大多数情况下, 几乎没有一个项目可\N以覆盖一亿人, 或产生相当的经济效益。
Dialogue: 1,0:02:01.14,0:02:02.41,Default,,0,0,0,,让我举个例子。
Dialogue: 1,0:02:03.92,0:02:09.32,Default,,0,0,0,,很多周末,我开车几分钟到附近的一家披萨店
Dialogue: 1,0:02:09.44,0:02:14.63,Default,,0,0,0,,买一块夏威夷披萨,这家店是一个绅士经营的。
Dialogue: 1,0:02:14.63,0:02:19.39,Default,,0,0,0,,他的披萨很棒,但是总是有一大堆披萨没卖掉变冷,
Dialogue: 1,0:02:19.53,0:02:22.31,Default,,0,0,0,,每个周末都会有几个口味的披萨缺货。
Dialogue: 1,0:02:23.29,0:02:25.99,Default,,0,0,0,,但是当我看他经营他的店铺时,
Dialogue: 1,0:02:26.7,0:02:32.13,Default,,0,0,0,,我感到兴奋,因为在他卖披萨的过程中,也产生了数据,
Dialogue: 1,0:02:32.23,0:02:38.16,Default,,0,0,0,,如果他能够使用人工智能,就可以从这些数据中获益。
Dialogue: 1,0:02:38.16,0:02:43.36,Default,,0,0,0,,如果输入了合适的数据, 人工智能系统很善于识别规律,
Dialogue: 1,0:02:43.44,0:02:49.10,Default,,0,0,0,,比如人工智能系统能识别出周五晚上地中海披萨卖得特别好,
Dialogue: 1,0:02:49.10,0:02:52.43,Default,,0,0,0,,这样就可以告诉老板周五下午多准备一些地中海披萨。
Dialogue: 1,0:02:53.46,0:02:58.26,Default,,0,0,0,,可能你会说: “嘿,Andrew, 这只是个小披萨店。 有什么了不起的?”
Dialogue: 1,0:02:58.26,0:03:01.59,Default,,0,0,0,,而我想说,对于店主来说,
Dialogue: 1,0:03:01.85,0:03:05.23,Default,,0,0,0,,如果可以帮他们每年多赚几千美元,
Dialogue: 1,0:03:05.37,0:03:08.95,Default,,0,0,0,,那对他们来说意义重大。
Dialogue: 1,0:03:08.95,0:03:14.56,Default,,0,0,0,,人们通常认为 AI 需要大量数据集才能很好的运行,
Dialogue: 1,0:03:14.56,0:03:15.96,Default,,0,0,0,,更多数据确实会有帮助。
Dialogue: 1,0:03:17.23,0:03:23.32,Default,,0,0,0,,但是如果没有大量数据,只有少量数据\N的情况下 AI 通常也能正常运行,
Dialogue: 1,0:03:23.46,0:03:26.84,Default,,0,0,0,,比如一家披萨店产生的数据量。
Dialogue: 1,0:03:26.84,0:03:30.92,Default,,0,0,0,,所以真正的问题不是来自披萨店的数据不足。
Dialogue: 1,0:03:30.92,0:03:34.86,Default,,0,0,0,,真正的问题是小披萨店无法为没有足够的客源,
Dialogue: 1,0:03:34.96,0:03:37.92,Default,,0,0,0,,利润不足以雇佣一个AI团队。
Dialogue: 1,0:03:39.55,0:03:44.43,Default,,0,0,0,,我知道在美国有大约五十万家独立餐厅,
Dialogue: 1,0:03:44.51,0:03:50.21,Default,,0,0,0,,这些餐厅总共为数以千万计的顾客提供服务,但每家餐厅都不同,
Dialogue: 1,0:03:50.37,0:03:53.81,Default,,0,0,0,,有不同的菜单、不同的顾客、不同的销售记录方式,
Dialogue: 1,0:03:53.95,0:03:56.99,Default,,0,0,0,,没有一个通用的 AI 系统能适用于所有餐厅。
Dialogue: 1,0:03:58.66,0:04:02.12,Default,,0,0,0,,如果我们能让小企业,
Dialogue: 1,0:04:02.20,0:04:05.44,Default,,0,0,0,,尤其是本地企业都能应用 AI 技术,会产生怎样的效果呢?
Dialogue: 1,0:04:05.44,0:04:10.91,Default,,0,0,0,,让我们来看看如果将 AI 应用于一家制造\N和销售T恤的公司可能会是什么样子。
Dialogue: 1,0:04:10.91,0:04:16.79,Default,,0,0,0,,我希望这家 T 恤公司的会计师可以使用 AI 预测需求。
Dialogue: 1,0:04:16.79,0:04:22.27,Default,,0,0,0,,比如,通过研究社交媒体上的热门话题,找\N出可以增加销量的好玩的表情包图案。
Dialogue: 1,0:04:23.37,0:04:25.43,Default,,0,0,0,,或者对于商品摆放,
Dialogue: 1,0:04:25.59,0:04:29.60,Default,,0,0,0,,为什么不可以让门店经理拍摄店面照片,
Dialogue: 1,0:04:29.68,0:04:34.90,Default,,0,0,0,,然后让 AI 推荐商品的摆放位置以提高销量呢?
Dialogue: 1,0:04:34.90,0:04:37.81,Default,,0,0,0,,对于供应链,能否让 AI 帮助买家货比三家,
Dialogue: 1,0:04:38.3,0:04:42.19,Default,,0,0,0,,向买家推荐这块 20 美元一码的布料已经是最低价,
Dialogue: 1,0:04:42.25,0:04:45.21,Default,,0,0,0,,还是有另外的店铺有更便宜的价格?
Dialogue: 1,0:04:46.34,0:04:47.86,Default,,0,0,0,,或者质量控制。
Dialogue: 1,0:04:47.86,0:04:59.35,Default,,0,0,0,,质检员应该能够使用 AI 自动扫描 T恤面\N料的图片,检查布料是否有破损或褪色。
Dialogue: 1,0:04:59.35,0:05:05.45,Default,,0,0,0,,如今,大科技公司借助 AI 已经可以很好\N的解决上面这些问题,并且效果显著。
Dialogue: 1,0:05:06.79,0:05:16.15,Default,,0,0,0,,但是对于传统的 T恤制造商、汽车修理厂、零售商、学校或本地农场,
Dialogue: 1,0:05:16.27,0:05:20.2,Default,,0,0,0,,当前还几乎没有 AI 的应用。
Dialogue: 1,0:05:20.2,0:05:24.24,Default,,0,0,0,,每家 T恤制造商都与其他家有很大的不同,
Dialogue: 1,0:05:24.38,0:05:27.22,Default,,0,0,0,,没有一个通用的 AI 系统能适用于所有厂商。
Dialogue: 1,0:05:28.34,0:05:33.34,Default,,0,0,0,,实际上,一旦你走出互联网和科技行业,
Dialogue: 1,0:05:33.42,0:05:35.85,Default,,0,0,0,,在其他行业,即使是大公司,
Dialogue: 1,0:05:36.3,0:05:39.81,Default,,0,0,0,,如制药公司、汽车制造商、医院,
Dialogue: 1,0:05:40.13,0:05:42.65,Default,,0,0,0,,也都面临着这个问题。
Dialogue: 1,0:05:42.65,0:05:45.31,Default,,0,0,0,,这就是人工智能的长尾问题。
Dialogue: 1,0:05:46.57,0:05:51.14,Default,,0,0,0,,如果你将所有当前和潜在的AI项目排序,
Dialogue: 1,0:05:51.22,0:05:55.28,Default,,0,0,0,,并按价值递减的顺序绘制图表,
Dialogue: 1,0:05:55.42,0:05:57.22,Default,,0,0,0,,你会得到一张像这样的图形。
Dialogue: 1,0:05:57.22,0:06:02.64,Default,,0,0,0,,也许最有价值的 AI 系统是决定在互联\N网上向人们展示哪些广告的系统。
Dialogue: 1,0:06:02.64,0:06:04.96,Default,,0,0,0,,也许第二有价值的是网络搜索引擎。
Dialogue: 1,0:06:04.96,0:06:08.46,Default,,0,0,0,,也许第三有价值的是一个在线购物商品推荐系统。
Dialogue: 1,0:06:09.57,0:06:12.35,Default,,0,0,0,,但当你走到这条曲线的右边时,
Dialogue: 1,0:06:12.55,0:06:15.67,Default,,0,0,0,,你会得到像 T恤商品摆放、
Dialogue: 1,0:06:15.75,0:06:20.25,Default,,0,0,0,,T恤需求预测或比萨店需求预测这样的项目。
Dialogue: 1,0:06:20.25,0:06:25.5,Default,,0,0,0,,每个项目都是独特的,需要定制开发。
Dialogue: 1,0:06:25.5,0:06:30.9,Default,,0,0,0,,同样是需求预测,对于 T恤的需求预测,\N它依赖于社交媒体上流行的表情包,
Dialogue: 1,0:06:30.17,0:06:36.39,Default,,0,0,0,,而比萨店的需求预测完全不同,它依赖于比萨店的销售数据。
Dialogue: 1,0:06:37.44,0:06:42.9,Default,,0,0,0,,如今成千上万的项目就处于这个分布的长尾上,
Dialogue: 1,0:06:42.21,0:06:46.65,Default,,0,0,0,,没有人去做这样的项目,但它们的总价值是巨大的。
Dialogue: 1,0:06:46.65,0:06:49.71,Default,,0,0,0,,那么我们该如何让小企业和个人,
Dialogue: 1,0:06:49.79,0:06:51.83,Default,,0,0,0,,有能力构建对他们有价值的 AI 系统呢?
Dialogue: 1,0:06:53.22,0:06:54.96,Default,,0,0,0,,在过去的几十年中,
Dialogue: 1,0:06:55.4,0:06:57.34,Default,,0,0,0,,如果你想构建一个 AI 系统,
Dialogue: 1,0:06:57.48,0:06:58.50,Default,,0,0,0,,需要做这些事:
Dialogue: 1,0:06:58.50,0:07:01.21,Default,,0,0,0,,你必须写很多行代码。
Dialogue: 1,0:07:01.21,0:07:03.99,Default,,0,0,0,,虽然我希望每个人都能学习编程,
Dialogue: 1,0:07:04.9,0:07:04.77,Default,,0,0,0,,事实上,
Dialogue: 1,0:07:05.7,0:07:07.11,Default,,0,0,0,,在线教育和线下教育正在帮助更多的人学习编程,
Dialogue: 1,0:07:07.19,0:07:12.86,Default,,0,0,0,,但不幸的是,并不是人人都有时间学习编程。
Dialogue: 1,0:07:13.87,0:07:17.63,Default,,0,0,0,,好在现在有一种新兴的构建 AI 系统的方式,
Dialogue: 1,0:07:17.75,0:07:20.13,Default,,0,0,0,,能让更多的人参与其中。
Dialogue: 1,0:07:20.13,0:07:29.78,Default,,0,0,0,,就像纸和笔是远比石板和凿子更先进的技术\N一样,在普及识字率方面起到了重要作用,
Dialogue: 1,0:07:29.92,0:07:32.82,Default,,0,0,0,,现在出现了新兴的 AI 开发平台,
Dialogue: 1,0:07:32.96,0:07:36.8,Default,,0,0,0,,不再需要写大量代码,
Dialogue: 1,0:07:36.24,0:07:39.30,Default,,0,0,0,,只需要你提供数据,
Dialogue: 1,0:07:39.36,0:07:41.98,Default,,0,0,0,,这对很多人来说更容易做到。
Dialogue: 1,0:07:43.27,0:07:47.88,Default,,0,0,0,,今天,有多家公司正在开发这样的平台。
Dialogue: 1,0:07:47.88,0:07:52.12,Default,,0,0,0,,让我用我们团队正在构建的一个平台来说明一些概念。
Dialogue: 1,0:07:52.12,0:07:58.52,Default,,0,0,0,,举例来说,检测员想要 AI 帮助检测布料上的瑕疵。
Dialogue: 1,0:07:58.52,0:08:00.60,Default,,0,0,0,,检测员可以拍下布料的照片,
Dialogue: 1,0:08:00.68,0:08:02.68,Default,,0,0,0,,上传到这样的平台上,
Dialogue: 1,0:08:03.56,0:08:09.43,Default,,0,0,0,,然后他们可以用矩形做标记,向 AI 展示布料裂缝长什么样。
Dialogue: 1,0:08:09.43,0:08:16.25,Default,,0,0,0,,他们也可以用矩形做标记,向 AI 展示布料褪色后长什么样。
Dialogue: 1,0:08:16.25,0:08:17.55,Default,,0,0,0,,这些图片
Dialogue: 1,0:08:18.3,0:08:21.42,Default,,0,0,0,,连同检测员标记的绿色和粉色矩形框
Dialogue: 1,0:08:21.52,0:08:27.12,Default,,0,0,0,,是检查员创建的数据,用来向 AI 展示如何检测裂缝和褪色。
Dialogue: 1,0:08:28.59,0:08:30.45,Default,,0,0,0,,在 AI 检查这些数据之后
Dialogue: 1,0:08:30.55,0:08:33.2,Default,,0,0,0,,我们会发现 AI 已经看过了足够的裂缝图片,
Dialogue: 1,0:08:33.12,0:08:35.52,Default,,0,0,0,,但还没有足够多的褪色图片。
Dialogue: 1,0:08:35.52,0:08:39.92,Default,,0,0,0,,这就好比一个初级检测员已经学会了如何准确地识别裂缝,
Dialogue: 1,0:08:40.6,0:08:43.70,Default,,0,0,0,,但是还得再磨练一下对褪色的判断。
Dialogue: 1,0:08:43.70,0:08:45.15,Default,,0,0,0,,所以检查员可以回去
Dialogue: 1,0:08:45.23,0:08:48.45,Default,,0,0,0,,并拍摄更多的褪色图片展示给AI
Dialogue: 1,0:08:48.57,0:08:50.73,Default,,0,0,0,,以帮助它加深对褪色布料的理解。
Dialogue: 1,0:08:50.73,0:08:55.15,Default,,0,0,0,,通过调整输入 AI 的数据,可以让 AI 变得更聪明。
Dialogue: 1,0:08:56.48,0:09:07.16,Default,,0,0,0,,借助这样容易操作的平台,再采购一套相\N应的摄影设备, 在几小时至几天内,
Dialogue: 1,0:09:07.26,0:09:16.71,Default,,0,0,0,,就能构建一个定制化的 AI 系统, 帮助工厂检\N测所有 T恤面料上的瑕疵、裂缝和褪色情况。
Dialogue: 1,0:09:16.71,0:09:24.12,Default,,0,0,0,,可能有人会说: “嘿,Andrew,这只是一家工厂, 有什么了不起的?”
Dialogue: 1,0:09:24.12,0:09:28.36,Default,,0,0,0,,我想告诉你,对于那些生活因此变得更\N轻松的检测员来说,这很有价值。
Dialogue: 1,0:09:28.36,0:09:35.47,Default,,0,0,0,,同样,这项技术可以让面包师借助 AI 检测他们制作的蛋糕质量,
Dialogue: 1,0:09:35.55,0:09:39.51,Default,,0,0,0,,让有机蔬菜农场主检测蔬菜的质量,
Dialogue: 1,0:09:39.61,0:09:42.85,Default,,0,0,0,,让家具制造商检测木材原料的质量。
Dialogue: 1,0:09:44.90,0:09:51.87,Default,,0,0,0,,这类平台也许还需要几年时间,才能将操作\N难度降低,能适用于每一个披萨店店主。
Dialogue: 1,0:09:51.87,0:09:54.7,Default,,0,0,0,,但是很多平台都在发展,
Dialogue: 1,0:09:54.15,0:10:00.4,Default,,0,0,0,,对于今天具备一定技术能力的人来说,有些平\N台只需要少量培训,就已经相当可用了。
Dialogue: 1,0:10:00.74,0:10:07.78,Default,,0,0,0,,这也意味着,我们不再依赖高级祭司来为其他人编写AI系统。
Dialogue: 1,0:10:07.88,0:10:16.19,Default,,0,0,0,,每位会计、 门店经理、 买家、检测员都有能力搭建自己的 AI 系统。
Dialogue: 1,0:10:17.55,0:10:25.1,Default,,0,0,0,,我希望比萨店老板和其他许多小企业主也能用上这项技术,
Dialogue: 1,0:10:25.47,0:10:30.94,Default,,0,0,0,,因为 AI 创造着巨大财富,也将在未来持续创造巨大财富。
Dialogue: 1,0:10:30.94,0:10:39.86,Default,,0,0,0,,只有让 AI 普及,我们才能确保 AI 创造\N的财富能被传播到社会的各个角落。
Dialogue: 1,0:10:39.86,0:10:47.68,Default,,0,0,0,,几百年前,我认为几乎没有人能够理解普及识字对社会的影响。
Dialogue: 1,0:10:47.68,0:10:55.11,Default,,0,0,0,,如今,我认为几乎没有人能够理解普及 AI 对社会的影响。
Dialogue: 1,0:10:55.11,0:10:58.95,Default,,0,0,0,,构建AI系统对大多数人来说几乎是不可能的,
Dialogue: 1,0:10:59.7,0:11:01.43,Default,,0,0,0,,但是未来不一定会是如此。
Dialogue: 1,0:11:01.43,0:11:03.33,Default,,0,0,0,,在即将到来的 AI 时代,
Dialogue: 1,0:11:03.49,0:11:07.11,Default,,0,0,0,,我们将赋予每个人为自己构建AI系统的能力。
Dialogue: 1,0:11:07.17,0:11:10.15,Default,,0,0,0,,我认为这将是一个非常令人兴奋的未来!
Dialogue: 1,0:11:10.15,0:11:10.83,Default,,0,0,0,,感谢大家!
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment