CCTV13 — broadcast
20261102 233000 UTC
233 transcript segments
Google Cloud Speech-to-Text API (Chirp) + Gemini 2.5 Flash Non-Thinking
Data courtesy of The GDELT Project (https://www.gdeltproject.org/), from the Internet Archive TV News Archive.
Machine transcription. Treat it as a searchable index of what was broadcast, not a verbatim quotation record.

[00:00:08] its mobile application has resumed operation, and Russian President's press secretary Dmitry Peskov said on the same day
[00:00:12] that the Russian President's website is subject to multiple cyber attacks every day,
[00:00:16] and electronic resources of various departments are always facing
[00:00:19] cyber attack threats
[00:00:21] .
[00:00:36] like stars helping dreams .
[00:00:40] CCTV National New Brand
[00:00:44] .
[00:00:48] Passionately striving to win with the world. Sailun Tires, sharing with you
[00:00:51] the moment of "Té" style.
[00:00:55] Sailun. Bao Bao Ke , Bao Bao Shou
[00:00:58] . Jin Zhenglai.
[00:01:31] So, is AI diagnosis reliable? Can it replace doctors? What is
[00:01:35] the medical capability of AI large models ? Some platforms also provide
[00:01:38] health consultation services for doctors' AI avatars. What kind of risks are behind
[00:01:42] this? Let's look at an example
[00:01:45] first.
[00:01:47] Not long ago, a parent took their three- year-old child to the First Affiliated Hospital of Zhejiang University of Chinese
[00:01:51] Medicine for pediatric emergency treatment. Dr. Li
[00:01:55] Lan considered that the child had an acute asthma attack complicated by lung infection and suggested
[00:01:59] that the child undergo
[00:02:03] further examination.
[00:02:08] He said, "Based on the results of this AI, I'm very sure that this child has an acute asthma attack." Then he said, "Just give me the prescription according to the treatment plan prompted by the AI."
[00:02:22] Afterwards, the parents took the child home, but a few days later, the parents brought the child back for a check-up.
[00:02:29] The result of AI often makes him question, he says, "How come what you said is different from what AI said?"
[00:02:48] So, can general AI large models really diagnose diseases directly? What are the medical capabilities of AI? With these questions, we came to the Shanghai Artificial Intelligence Medical Intelligence Center to
[00:03:04] test several general large models. The Shanghai Artificial Intelligence Laboratory Medical Intelligence Center has established the first
[00:03:08] domestic platform for medical large model application evaluation and verification
[00:03:12] . The staff here told reporters that they have specially designed
[00:03:15] a set of standardized test questions covering 36 screening
[00:03:19] levels and 1 million test questions , which can
[00:03:23] evaluate the medical application capabilities of large models from multiple dimensions such as medical knowledge Q&A, complex medical reasoning, medical safety
[00:03:27] and ethics .
[00:04:44] story in
[00:05:06] Seven real clinical cases were selected, and five general large language models were asked to interpret them separately. Then, more than 400 professional doctors of different ages and regions were invited to
[00:05:10] evaluate their interpretations. At the same time, researchers also
[00:05:14] configured 421 AI intelligence bodies simulating different doctor backgrounds
[00:05:18] , asking them to rate the same answers.
[00:05:21] The study found that AI intelligence bodies and human doctors
[00:05:25] have certain similarities, but in complex detail processing,
[00:05:29] AI intelligence bodies are still unable to completely replace human doctors
[00:05:32] . The relevant research results were published in the digital medicine
[00:05:36] top journal NPJ Digital Medicine
[00:05:40] under Nature. Everyone can try it themselves, for example, if we say.
[00:05:42] 35 years old
[00:05:44] This is one input method. There is another input method: 35-year-old male with
[00:05:48] abdominal pain for two weeks
[00:05:49] .
[00:05:55] The conclusions given by the two input methods must be different .
[00:05:58] 。
[00:06:05] may not be professional enough or comprehensive enough, or some information that we consider very important
[00:06:09] may be overlooked by it . Experts say that
[00:06:14] the answers given are highly dependent on the descriptive information provided by the user. However, in medical
[00:06:17] scenarios, due to the lack of medical background, ordinary patients
[00:06:21] often find it difficult to accurately describe their condition. Misinformation
[00:06:25] is very common in actual consultations. This
[00:06:28] information expression error can be gradually grasped by doctors during consultations through
[00:06:32] observation, listening, questioning, and pulse diagnosis. However, users use online
[00:06:36] AI consultations mostly through screen-based input and output. This
[00:06:40] passive text reception will directly affect the application scenarios of AI are constantly expanding, and AI medical products are constantly emerging. Some platforms have provided doctors'
[00:07:32] AI avatar health services.
[00:07:35] What kind of risks are behind these digital doctor applications?
[00:07:55] The investigation found that in addition to Q&A AI giving medical advice based on the patient's description, in many apps, a large number of digital doctors, under the name of a famous doctor or a famous doctor's team, have appeared one after another. They directly use the names of certain experts
[00:07:58] to provide health consultations for users in the software
[00:08:02] . However, when reporters contacted these doctors themselves, they found that
[00:08:06] some doctors were not aware that they also had a digital
[00:08:10] avatar on the app. In addition, these experts, AI-generated suggestions are given, but in the actual process, users may directly skip this step, then the doctor,
[00:08:42] then he may blame the doctor at that hospital. The current problem
[00:08:46] is a question of defining responsibility. For example, if I use
[00:08:50] artificial intelligence for auxiliary diagnosis, then if there is a wrong diagnosis, whose responsibility
[00:08:54] is it? One is that legislation is very important , for the government to do
[00:08:57] this. Another is to test and evaluate
[00:09:01] the level of the artificial intelligence that we are currently using
[00:09:05] .
[00:09:08] Last year, the National Health Commission and other departments issued guidelines on promoting and regulating
[00:09:42] This market is rapidly booming. Based on current observations, China has
[00:09:46] more than 100,000 newly registered enterprises in the AI digital healthcare sector
[00:09:50] . With the rise of such market entities , their services, products
[00:09:54] , and even the ecosystem of software and hardware integration will rapidly emerge
[00:09:57] . It is understood that the application of artificial intelligence in the medical field has actually
[00:10:01] been widely popularized for a long time. As early as 10 years ago, auxiliary diagnoses in
[00:10:05] endoscopy, CT and other fields have already been introduced into medical equipment
[00:10:09] . This year, the First Affiliated Hospital of Zhejiang University School of Medicine and other
[00:10:13] medical institutions collaborated with artificial intelligence enterprises to launch the world's first
[00:10:17] expert-level general medical imaging model. The relevant research
[00:10:20] results were published in the international top academic journal
[00:10:24] Science
[00:10:30] . However, it is undeniable that patients directly using generative large models
[00:10:34] for self-diagnosis and treatment still carry significant risks. I think
[00:10:38] in the near future, there will definitely be a standard for everyone to work together to
[00:10:41] make a set of guidelines. This guideline has two aspects: one is the bottom line, meaning
[00:10:45] what lines you cannot cross, and then what is the room for improvement. I think
[00:10:49] there should be such a setting. First, it's about access, and second, it's likely to be dynamic. AI products, the application potential of artificial intelligence in the medical field is gradually being realized, but medical power deals with human health and well-being. Technical answers
[00:10:53] 可 能 是 动 态 的 。 AI品,人工能在醫料領域的應用前力正在一步步的轉化為現實,但是醫權面對的是人的健康與,技術上的回答
[00:11:25] do not represent clinical diagnostic capabilities , let alone mean that
[00:11:29] they can independently assume diagnostic responsibility . What can be handed over to AI, and what
[00:11:33] decisions must be made by doctors ? And if something goes wrong,
[00:11:36] who should be responsible? Answering these questions not only relates to the safety
[00:11:40] of every patient, but also to the sound development of the artificial intelligence medical
[00:11:44] and health industry. Clearly define the red line between chatting and consultation,
[00:11:48] let technological progress and safety assurance synchronize, allow innovation
[00:11:52] to have space, application to have boundaries , and responsibility to be regulated
[00:14:00] Let's learn more through a short video. The US midterm elections are elections held in the middle of the president's four-year term
[00:14:03] . In this year's midterm elections , the Republican and Democratic parties will compete
[00:14:07] for all 435 seats in the House of Representatives and
[00:14:11] 35 of the 100 seats in the Senate . In addition, voters
[00:14:14] will also elect governors of 36 states and local public officials such as
[00:14:18] state legislators in some states . Midterm elections
[00:14:22] have always been regarded as a re-election test for the incumbent president
[00:14:26] . The president's party often cannot escape the "midterm election curse" of losing power
[00:14:30] . Data shows that in the four midterm elections since 2010
[00:14:34] , the president's party has lost seats in the House of Representatives
[00:14:38] and lost seats twice in the Senate
[00:14:41] . Political scholars believe that this is partly due to voters expressing
[00:14:45] dissatisfaction with the ruling party through midterm elections , and partly influenced by
[00:14:49] factors such as the mobilization capabilities of the two parties and voter enthusiasm
[00:14:52] . In short,
[00:15:07] but the advantage is not great. Mainstream US election forecasts show that the Democratic Party is very likely to
[00:15:11] win the House of Representatives in this midterm election, and still has a certain chance to
[00:15:15] flip the Senate. Next, let's learn about the US Congress's House and Senate
[00:15:18] and the power of the Democratic and Republican parties in Congress
[00:15:22] through a short video .
[00:15:25] The US Congress is the highest legislative body in the United States, composed of the Senate and
[00:15:29] the House of Representatives. The House of Representatives has legislative agenda, budget expenditures
[00:15:33] and oversight powers , with a total of 435 seats
[00:15:38] . Among them, the Republican Party holds 218 to 21 seats,
[00:15:41] and the Democratic Party holds 214 seats . The Republican Party holds a slim
[00:15:45] majority. The Senate, however, has the power to confirm important personnel appointments
[00:15:49] such as federal judges and cabinet officials , with a total of 100 seats
[00:15:53] . Among them, the Republican Party holds 53 seats, and the Democratic Party
[00:15:57] holds 45 seats. The other two seats are held by independents, but they usually
[00:16:01] cooperate with the Democratic Party caucus . In this midterm election
[00:16:05] , the Republican and Democratic parties will compete .
[00:17:07] Scene 1: Defection of a legislator. On November 1,
[00:17:11] US Michigan Democratic Federal Representative Haley Stevens announced
[00:17:15] that she was withdrawing her support for the state's Democratic Federal Senator
[00:17:19] candidate Abdul El-Sayed, and instead supporting the state's Republican Federal Senator
[00:17:23] candidate Mike Rogers. Stevens emphasized that she has not abandoned the Democratic
[00:17:26] Party, but merely chose to support the Republican Party
[00:17:30] in this election
[00:17:32] .
[00:17:40] Sayed said later that Stevens' actions were insignificant, what was important
[00:17:44] was that Michigan residents were suffering from the cost of living crisis
[00:17:47] . Reuters reported that Stevens' move was an unexpected boost
[00:17:51] for the Republican Party . Many Democrats publicly
[00:17:55] condemned Stevens. The Michigan Democratic Party chairman announced that he would
[00:17:58] kick her out of the party, while US President Trump praised
[00:18:02] Stevens for being unprecedentedly loyal to Michigan .
[00:18:06] State
[00:18:59] . And these differences will continue to exist after the midterm elections. Scene 2: Mother and daughter turn against each other. In the Ohio gubernatorial election,
[00:19:03] Republican candidate Vivek Ramaswamy challenged the Democratic candidate
[00:19:07] .
[00:20:07] now that I had a pretty rough childhood growing up and now my opponent
[00:20:11] has chosen to exploit
[00:20:14] intentionally this
[00:20:21] a desperate attempt to save his struggling
[00:20:25] campaign. this is so sick and
[00:20:28] depraved. After the broadcast, it was widely criticized by various sources, including the chairman, who believed that
[00:22:20] Data shows that the day-ahead electricity spot prices in Tokyo and central Japan
[00:22:24] were 36.11 yen per kilowatt-hour (approximately 1.52 yuan RMB)
[00:22:28] and 36.29 yen (approximately 1.53 yuan RMB), respectively,
[00:22:31] an increase of 1.1% and 7.2% respectively from a week ago,
[00:22:35] reaching a new high since September 2022
[00:22:39] . Data from the Japanese Agency for Natural Resources and Energy shows that power generation from
[00:22:43] liquefied natural gas-fired thermal power accounts for about 30%
[00:22:47] of Japan's total power generation . Therefore, electricity prices are easily affected
[00:22:50] by fluctuations in the international liquefied natural gas market
[00:22:54] . According to Japanese media reports, seven of Japan's ten major power companies
[00:22:58] announced an increase in residential electricity prices in November, with prices reaching a new
[00:23:02] high since records began .
[00:23:06] Recently, domestic prices in Japan have experienced a new round of price hikes, increasing the burden on people's
[00:23:10] daily lives . A survey by Japan's Teikoku Databank company
[00:23:13] shows that nearly 1,500 kinds of food will increase in price in Japan in November,
[00:23:17] and the total number of food items expected to increase in price this year will exceed
[00:23:21] 20,000 . In addition, gas
[00:23:25] fees, including basic fees and expressway tolls, have also increased
[00:23:28] .
[00:23:43] next two days, Jianghuai, Jianghan, western and
[00:23:46] northern Jiangnan, as well as Chongqing and eastern Guizhou.
[00:25:20] Please introduce it to us again. Okay, actually, the warmth during the alternating seasons of autumn and winter
[00:25:23] is indeed not reliable. It is expected that in the latter half of this week, the cold air affecting
[00:25:27] our country will become very frequent . Affected by this, in the next few days, the maximum temperature
[00:25:31] in Northeast China, North China, and other places will drop by about 10 degrees. After the temperature drop
[00:25:35] , the maximum temperature will widely set a new low since the second half of this year
[00:25:39] . The maximum temperature in many places in North China will drop to single digits. Even
[00:25:43] the middle and lower reaches of the Yangtze River will drop to around 15 degrees
[00:25:46] . The weather has clearly turned cooler , so please pay attention.
[00:26:31] It is understood that Budong Reservoir mainly focuses on flood control and irrigation. These two days , the local air humidity is high
[00:26:34] , and fog is frequent, making the autumn scenery of
[00:26:45] Beijing, clear to cloudy, 6 to 19 degrees. Yichun
[00:26:49] , cloudy turning to light snow, minus 10 to 0 degrees
[00:26:51] . Hulunbuir , light snow turning clear, minus 12 to minus
[00:26:55] 7 degrees. Ordos , clear turning cloudy, 3 to
[00:26:59] 14 degrees. Yinchuan, cloudy.
[00:27:05] 3 to 16 degrees. Lanzhou, overcast, 3 to 14 degrees
[00:27:09] . Lhasa, clear, 1 to 17 degrees
[00:27:12] . Dazhou, light rain turning overcast, 14 to 16
[00:27:15] degrees . Ya'an, light rain, 13 to 18
[00:27:19] degrees . Zunyi, light rain, 13 to 15
[00:27:23] degrees .
[00:27:27] to 26 degrees
[00:27:29] .
[00:27:46] Weifang overcast 11 to 12 degrees
[00:27:51] Qingdao 17 to 20 degrees
[00:27:55] Sanmenxia light rain 8 degrees
[00:27:59] Hefei light rain turning moderate 14 to 19 degrees
[00:28:04] Huai'an, overcast turning light rain, 14 to 20 degrees.
[00:28:07] Yancheng, overcast turning light rain, 15 to 20 degrees.
[00:28:11] Taipei, light rain turning partly cloudy, 21 to 24 degrees
[00:28:15] . Hong Kong, partly cloudy, 25 to 30 degrees
[00:28:19] . Macau, clear turning partly cloudy, 25 to 30
[00:28:21] degrees
[00:28:25] . Yifang Qinghong. One side water, one side
[00:28:28] people .
[00:28:35] Looking at the market for thousands of years, the bright white
[00:28:38] clouds by the lake. Hubei
[00:28:42] by the white clouds, wine night. The beauty of mountains and rivers
[00:28:45] , magical moving village, the soul of Danxia, a thousand-year-old spiritual song
[00:28:49] , overcoming divine power , romantic Wuzhou
[00:28:53] . Dream ancient formation, the most beautiful in the world, bumping
[00:28:57] into media Guangxi .
[00:29:00] 。
[00:29:32] year Yuanjiang . Every
[00:29:35] thread, every step, is made into a new pattern of development.
[00:29:39] Every speech, every gathering, ignites the power of shared progress
[00:29:43] . Move forward with determination
[00:29:46] . China Construction Bank
[00:29:50] . Collecting ancient scriptures, connecting the world, calming
[00:29:54] the waves, sailing the world
[00:29:57] .
[00:30:52] news .
[00:30:55] On November 2nd, Xi Jinping held a ceremony in the North Hall of the Great Hall of the People
[00:30:58] .