Morning Post | Apple may say goodbye to "Hey Siri"/Microsoft terminates support for Cortana/Ted Jiang says AI has no intelligence

爱faner morning report reading

• WWDC will set up an Apple head-mounted display experience area
• Apple’s voice assistant may have a major upgrade and may say goodbye to “Hey Siri”
• Provide real tactile feedback, "artificial skin" technology comes out
• "Farewell Cortana", Microsoft announced that it will end support for the assistant Cortana on Windows
• Report shows that generative AI revenue is expected to reach $130 million in 2032
• Sony CEO will use AI to enhance cloud gaming experience in the future
• The world’s most popular online programming course, CS50 plans to introduce AI teaching assistants
• The return capsule of the Shenzhou 15th manned spacecraft successfully landed
• Ted Jiang talks about AI: "AI will not seriously prepare a joke for you"
• Tesla Model S Plaid Track Pack breaks record for mass-produced electric vehicles
• Huawei “Pangu Chat” was released on July 7, targeting government and enterprise users
• Google next-generation processor Tensor G3 specifications leaked
• Domestic flight fuel surcharge will be reduced by 10 yuan and 30 yuan starting today
• Beijing issues a new round of consumer coupons, each person can receive up to 1,600 yuan
• Hayao Miyazaki's last work, "What Kind of Life Do You Want to Live" by Zero Ann Fat will be released on July 14th
• "Spider-Man: Across the Universe" North American first-day box office set annual record

The return capsule of the Shenzhou 15th manned spacecraft successfully landed, and the Shenzhou 15th manned mission was a complete success

According to the China Manned Space Engineering Office, at 6:33 on June 4, 2023, Beijing time, the return capsule of the Shenzhou 15 manned spacecraft successfully landed at the Dongfeng Landing Field. The on-site medical supervision and medical insurance personnel confirmed that the astronauts paid Junlong, Deng Qingming, and Zhang Lu are in good health, and the Shenzhou 15 manned mission was a complete success.

Successful separation! The orbital module and return module of Shenzhou 15 were successfully separated at 5:42 according to flight procedures. After that, the spacecraft returned to brake and ignited the engine, and the return capsule separated from the propulsion module. The return capsule landed successfully. The search and rescue team responsible for the search, rescue and recovery mission found the target in time and arrived at the landing site. After the door of the return module was opened, medical supervision and insurance personnel confirmed that the astronauts were in good health.

The Shenzhou 15 manned spacecraft was launched from the Jiuquan Satellite Launch Center on November 29, 2022, and then docked with the Tianhe core module to form a combination. During their stay in orbit, the three astronauts completed a large number of space science experiments, conducted 4 extravehicular activities, and successfully completed the installation of the extravehicular expansion pump unit, the installation and connection of cross-cabin cables, and the extravehicular load exposure platform. Support rod installation and other tasks, coupled with the completion of multiple space station cargo evacuation missions, laid the foundation for subsequent large-scale extravehicular science and technology experiments.

As the astronaut crew with the oldest average age when performing missions so far, they not only broke the record for the number of extravehicular activities of a single Chinese astronaut crew, but also witnessed the historical moment when the Chinese space station was fully completed.

Apple’s voice assistant may have a major upgrade and may say goodbye to “Hey Siri”

In the latest set of revelations, Mark Gurman said that Apple may announce major changes to Siri at WWDC. The change will no longer use the phrase "Hey Siri" to trigger Apple's voice assistant, but only need to say "Siri ”.

Mark Gurman said that Apple is upgrading its existing voice assistant so that Siri can understand and respond to commands more directly. Such technology conversion may seem like a small move, but it is a big technical challenge and requires a lot of manual labor. Smart training and groundwork.

"The complexity is reflected in the fact that Siri needs to be able to understand the single word "Siri" in multiple accents and dialects, while the two-word "Hey Siri" is more likely to increase the possibility of the system accepting the command correctly."

Gurman also said that Apple is now committed to allowing more third-party applications and services to integrate Siri to provide better assistance, and this series of voice assistant upgrades will be launched sometime in 2023-2024, because Apple's latest headset will feature voice control, so this major upgrade to Siri may also be announced at the conference.

Ted Jiang talks about AI: "AI will not seriously prepare a joke for you"

The 55-year-old well-known science fiction writer Jiang Fengnan (Ted Jiang, Ted Chiang recently talked with a reporter from the Financial Times about the current AI trends and the relationship between science fiction and it.

"The machines we have now, they have no consciousness," Jiang said. "When one person teaches another person, that is the interaction between consciousness."

Ted Chiang believes that, at the same time, artificial intelligence models are trained by switching the so-called "weights" or the strength of the connections between different variables in the model to obtain the desired results.

"It's like when you think you're teaching a child and all you're doing is adjusting the weights in the network, that would be a real mistake."

What Jiang objects to is that current artificial intelligence practitioners and public opinion talk about large language model training allowing AI to "learn", "understand" and "know". This series of anthropomorphic languages ​​creates an illusion for us. . Even experienced practitioners can easily be fooled into thinking that AI is creating a spark of perception when in fact it is not.

Jiang quoted from a recent exchange with netizens on Twitter, talking about a doubt as to why we are still using a word "artificial intelligence" from the 1950s to describe today's new scientific development trends. He believed that a new term must be invented. , "Applied Statistics", he believes that the answers given by current AI are actually statistically analyzed from a large amount of text. In his opinion, this does not make the tool intelligent, "but people are obviously unwilling to use this Word, because this word is not interesting enough."

As a science fiction writer, Ted Chiang pays special attention to the two themes of "language" and "intelligence". He is concerned about the way language shapes our way of thinking and self-perception, and whether free will exists.

When a reporter asked him about his views on using ChatGPT for novel creation, Ted Jiang said that large language models are the basic technology of chatbots such as ChatGPT and Google Bard, but this technology is mainly used to generate text that no one necessarily wants to read. Or writing "filler text," a task that anthropologist David Graeber calls "bullshit work," he admits that AI-generated novels or texts are not enjoyable, but that they may be useful in certain areas.

In a widely circulated article published in The New Yorker in February this year, Jiang also believed that language models use text for training and vague imitation, which is actually rearranging word sequences that comply with grammatical rules. Because the technology generates content that is slightly different from existing ones, it gives the impression of understanding.

"Your daughter told jokes and found them funny. There are strong social factors behind this, but ChatGPT did not find anything funny and did not want to be funny." He believed that ChatGPT was not "rehearsing in the mind, See if it makes you laugh the next time you hang out."

Without humans giving AI intentions, emotions, and goals, the language it uses is meaningless. "Language is a way to facilitate interaction with other living things. This is completely different from the 'tag prediction' used to train AI today."

Jiang is also worried that “AI will become a ‘power amplifier’ for capitalism” and will let capital pursue their goals as much as possible, but turn a blind eye to the possible negative consequences.

At the end of the interview, Ted Chiang cited a volleyball named "Wilson" played by Hanks in Tom Hanks's classic movie "Cast Away". At that time, he was living alone on a desert island and drew pictures on the volleyball. He cut his face and spoke to him, which encouraged him to finally escape. Jiang believes that today’s AI chatbots are actually the same as “Wilson”:

"People find it difficult to interact with people, there are too many complexities, but AI does not need your return. Modern life often makes people feel like you are living on a desert island, and you long for companionship. Because of this, AI chatbots It can become popular quickly because it provides people with the same comfort as "Wilson"."

"We always hope that the other end of our conversation is a human being, but in fact there is not."

WWDC will set up an Apple head-mounted display experience area

Apple will plan to offer some developers and attendees the opportunity to try out the new headset at its upcoming WWDC 2023 event, according to Bloomberg’s Mark Gurman.

According to reports, Apple Park, Apple’s headquarters, is building a new “facility” designed to provide a better head-mounted display experience. The testing area is located on the basketball court at Apple Park, near a fitness center that employees can use.

It is reported that this experience area will only invite some media personnel and developers to give demonstrations, but the entire experience area will be retained after WWDC to continue to invite developers to experience the new headset based on the development of applications.

It is worth noting that the new Apple headset will not be released immediately after WWDC, but will be launched in a few months. Mass production has not yet begun, and the latest news shows that it will be launched before December.

"Farewell Cortana", Microsoft announced that it will end support for Windows Cortana

With the advent of Microsoft's artificial intelligence tool Windows Copilot, Microsoft recently announced that it will terminate updates and support for Windows Cortana later this year.

At the Microsoft Build conference in May, a new Windows Copilot tool integrated with artificial intelligence will be released. This tool will exist in your taskbar and use artificial intelligence to complete everything that Cortana can do before. Even Copilot can Summarize content, rewrite text, ask questions, adjust computer settings, etc.

Cortana was officially launched at the 2014 Microsoft Developer Conference. As an intelligent personal assistant developed by Microsoft, it is named after the female-shaped artificial intelligence in Microsoft's well-known FPS game "Halo". It is also the most important part of the work. One of the characters.

Initially, the official did not intend to name this artificial intelligence after this, but only used Cortana as the development code name. However, later, due to a large-scale petition from users, the name became the official name.

"Cortana" released in 2014 is known as the world's first cross-platform intelligent personal assistant, and the Chinese version is named after it. However, in recent years, "Cortana" has had a troubled life. First, in early 2021, the Cortana application on iOS and Android has ceased service. Now, with the advent of Copilot, Cortana will officially enter history.

Report shows that generative artificial intelligence revenue is expected to reach $130 million in 2032

Generative artificial intelligence could generate $1.3 trillion in revenue by 2023, up from a current market size of just $40 billion, according to a new report from Bloomberg Intelligence.

The report pointed out that data shows that with the continuous improvement of artificial intelligence infrastructure in the near future, growth may expand at a compound annual growth rate of 42%, and gradually develop towards large-scale language models, digital advertising, professional software and services, etc.

Additionally, growing demand for generative AI products, driven by technologies such as AI assistants, new AI infrastructure, and AI programming assistants, could drive approximately $280 billion in software revenue. As more businesses move their work to the cloud, companies like Amazon, Microsoft, Google and Nvidia could be the biggest beneficiaries.

By 2032, AI-related spending will account for 10% of the total spending on IT hardware, software services, advertising spending and gaming markets, compared with less than 1% now. The main driver will be generative AI infrastructure as a Services for training large language models ($247 billion by 2032), followed by digital advertising powered by the technology ($192 billion) and specialized generative AI assistant software ($89 billion).

On the hardware side, revenue will be driven by AI servers ($132 billion), AI storage ($93 billion), computer vision AI products ($61 billion) and conversational AI devices ($108 billion).

Generative artificial intelligence may bring benefits in the life sciences and education fields, which deserve special attention. Analysis by Bloomberg Intelligence found that early use cases for ChatGPT suggest both areas may be growing rapidly within the larger software landscape. Professional software assistants using artificial intelligence technology are revolutionizing search and information aggregation methods, which will further drive the development of both market segments.

Provide real tactile feedback, "artificial skin" technology is launched

Recently, scientists at Stanford University have developed a soft and stretchable "artificial skin" that can directly connect tactile feedback to the brain, thereby simulating the sensory feedback of real skin. If this technology can be improved , will benefit millions of prosthetic limb users.

This research article published in the magazine "Science" in early May introduces this latest academic achievement. The researchers said, "We are inspired by the natural system and want to imitate it. Maybe one day we can help patients not only return to sports function, and can also restore their sense of touch."

Scientists have long dreamed of creating prosthetics that not only restore movement but also provide sensations (for example, sensing pressure, temperature and vibration) to help patients return to a normal life. Skin injuries and amputations can wreak havoc on perception and movement, so even something as simple as feeling the coolness of a cold beer and grabbing an object can be challenging.

Current experiments have made huge strides in mice, where the animal responded when researchers touched its "artificial skin" and sent electrical pulses to its brain.

Using "artificial skin", robots can feel like humans so that they can be more flexible at work. This is critical for safety in industries that require physical interaction between humans and robots, such as in some manufacturing and construction industries, such as robots passing tools to humans on the manufacturing floor.

But touch is complicated. Human skin has millions of receptors, and when they are poked, pressed, squeezed, or burned, they respond by sending electrical impulses through nerves to the brain. The brain sends information back to tell the muscles to move.

This latest "artificial skin" has a stretchable organic transistor network layer that can receive and transmit electrical signals. These layers are only about 25 to 50 microns thick, and when sandwiched in between, are as thin as a piece of paper, similar to skin. Designed to sense pressure, temperature, strain and chemicals. They were able to convert this sensory information into electrical signals and manipulated them with just 5 volts.

However, these technologies still have flaws. Researchers say that human skin understands the feeling of an object and can then predict it, but "artificial skin" does not have such a memory; another challenge is that the current signal transmission is still too slow and Unable to be used in practice, such a delay "won't allow us to get a real feel for it".

Currently, researchers are still further advancing this research, aiming to install more different sensors into electronic skin to more closely replicate the tactile sensation of human hands.

Sony CEO will use AI to enhance cloud gaming experience in the future

Sony Group CEO Kenichiro Yoshida said in an interview with the Financial Times that the company is still paying attention to the development of cloud games and may use artificial intelligence to enhance the experience of cloud games in the future.

Kenichiro Yoshida believes that current cloud games still face many problems, including network latency, server costs, etc. He feels that most users now still use game consoles or computers to play games, rather than streaming cloud games.

Following the failure of OnLive and Google's Stadia, while Microsoft's Game Pass subscription service is growing, Kenichiro Yoshida revealed that Sony may use its artificial intelligence agent Sophy for cloud gaming in the future, and has already taken advantage of the low price of games. Rush hour to train Sophy on how to beat human players in the racing game Gran Turismo.

The world’s most popular online programming course, CS50 plans to introduce AI teaching assistants

Harvard University’s CS50 course, known as “the world’s most popular online programming learning course,” is preparing to introduce AI teaching assistants, which will use AI to assist students in programming learning to further lower the threshold and improve efficiency.

Professor Malan, 46, said that the current CS50 has hundreds of on-campus students and more than 40,000 online students from different time zones with different levels of knowledge and experience. Even if there are more than 100 teaching assistants during the class, it is difficult Answer all questions completely.

So Professor Malan’s team introduced ChatGPT-style generative AI to grade students’ homework, evaluate students’ programming level and give feedback. This process is not a simple error correction and correction, but a real process. Provide advice and help to students.

Tesla Model S Plaid Track Pack sets a new lap record for mass-produced electric cars at the Nürburgring

A Tesla Model S Plaid equipped with the Track Pack kit set a new record for the fastest lap of a mass-produced electric vehicle at the Nürburgring in Germany yesterday, 8 times faster than the previous record of the Porsche Taycan Turbo S More than seconds.

This Model S Plaid’s lap time of 7 minutes 25 seconds 231 milliseconds (7:25.231) is second only to supercars such as the Ferrari Enzo and the Mercedes-Benz SLS AMG Black Series. The Porsche Taycan Turbo S broke the Tesla Model S Plaid's fastest electric vehicle lap record set a year ago with a time of 7 minutes and 33 seconds in August last year. Today, Tesla has regained that honor.

Tesla Model S Plaid is currently one of the fastest mass-produced electric vehicles in the world, with the ability to break 100 km/h in 2.1 seconds. Its Track Pack package can also unlock the car’s top speed of 322km/h.

Huawei “Pangu Chat” was released on July 7, targeting government and enterprise users

According to TMTpost Media’s exclusive knowledge from within Huawei, Huawei will release a multi-modal 100 billion-level mobile phone called “Pangu Chat” at the Huawei Cloud Developer Conference (HDC.Cloud 2023) on July 7 this year. Large model products directly compete with ChatGPT and are released for external release and internal testing. This product is mainly aimed at government and enterprise customers, including the To B/G field.

The Huawei Pangu large model was originally proposed internally in Huawei Cloud in November 2020 and officially released to the public in April 2021. According to a paper published by Huawei, the number of parameters of Huawei's PanGu-Σ large model is as high as 1.085 trillion, and it is developed based on Huawei's independently developed MindSpore framework. Overall, the performance of the PanGu-Σ large model in dialogue may be close to the level of GPT-3.5.

Subsequently, an upgrade to version 2.0 was carried out in April 2022. Currently, among the large AI models, the natural language processing (NLP) large model, the computer vision (CV) large model, and the scientific computing (meteorology) large model have been marked as being about to go online.

According to reports, Huawei Pangu is the first Chinese pre-trained large-scale model with 1 billion parameters, and the CV large-scale model has also reached the scale of 3 billion parameters for the first time. The Pangu CV large model is the largest CV large model in the industry. It achieves both discriminative and generative capabilities and leads the industry in small sample learning on ImageNet. The Pangu meteorological model can provide second-level weather forecasts, while Zidong. Taichu is the world's first large-scale model that supports three modes: image, text and audio.

Google next-generation processor Tensor G3 specifications leaked

According to Google’s internal information obtained by Android Authority, Google’s next-generation processor Tensor G3 will use a unique 9-core CPU architecture, and the GPU will add ray tracing capabilities.

The CPU of Tensor G3 will be composed of 1 Cortex-X3 core at 3.0GHz, 4 Cortex-A715 cores at 2.45GHz, and 4 Cortex-A510 cores at 2.15GHz. Compared with previous generations, the frequency has been increased, which may bring Here comes some performance improvements. There may still be a considerable gap compared to the latest released Arm v9.2 architecture.

Google officially launched its first customized SoC Tensor for smartphones two years ago and used it on its Pixel phone. It is reported that this Tensor G3 may be used in the latest Pixel 8 in the series.

Domestic flight fuel surcharge will be reduced by 10 yuan and 30 yuan starting today

Many airlines announced that the fuel surcharge standards for domestic flights will be adjusted starting from June 5. This is the third reduction this year.

According to the new standards, adult passengers will be charged a fuel surcharge of 20 yuan per passenger on routes below 800 kilometers, and 30 yuan on routes above 800 kilometers. Compared with the previous standards, they were reduced by 10 yuan and 30 yuan respectively.

Beijing issues a new round of consumer coupons, each person can receive up to 1,600 yuan

In order to accelerate the release of green consumption potential and expand service supply at important nodes such as the Dragon Boat Festival and summer vacation, Beijing will issue a new round of "Beijing Lottery Green" consumer coupons from June 3 to July 16. Consumers in Beijing will A person can receive up to five consumption coupons worth a total of 1,600 yuan.

According to reports, consumers can receive coupons through the online platforms of 19 companies including JD.com, Lenovo, Xiaomi, Suning, and Red Star Macalline starting from 10:00 a.m. every day during this round of consumption coupon issuance.

Consumers can receive one or more, up to five, totaling 1,600 yuan on the same enterprise online platform, which can be used to purchase mobile phones, computers, bicycles, monitors, air purifiers, smart sofas, Smart beds, dishwashers, TVs, air conditioners, refrigerators, washing machines, printers, projectors, sweeping robots, vacuum cleaners, floor scrubbers, learning machines, speakers, smart door locks, bracelets, smart watches, smart glasses, etc. 55 There are more than 20,000 products in major categories.

Hayao Miyazaki's last work, "What Kind of Life Do You Want to Live" by Zero Ann Fat will be released on July 14

Hailed by Studio Ghibli President Koji Hoshino as "Ghibli's largest production to date", "What Kind of Life Do You Want to Live" (君たちはどう生きるか) is the career of the famous animation director Hayao Miyazaki. The last work in. The movie will be released in Japan on July 14, but no stills or trailer have been released so far except for a poster.

Reports say that Hayao Miyazaki and Studio Ghibli adopted a "zero publicity" strategy, not announcing in advance and not releasing film stills. The film is adapted from the work of the same name published by novelist Genzaburo Yoshino in 1937. It tells the story of a 15-year-old boy named "Benny" who, under the inspiration and guidance of his uncle, a college graduate, explores friendship, wealth, poverty, discrimination, bullying, Issues such as people and society, courage, etc. aim to clarify "the value of what makes people human."

Hayao Miyazaki originally announced his retirement after the premiere of the 2013 movie "The Wind Rises", but out of the idea of ​​leaving a work for his grandson, he announced his comeback. He expressed his hope to tell his grandson through this movie: "Grandpa is about to go to another world, but he left this movie for you."

"How Do You Want to Live" has been in production since Hayao Miyazaki announced his return at the end of October 2017. It was originally planned to be released in 2019, but due to delays in the production schedule, it is finally scheduled to be released in July this year.

"Spider-Man: Across the Universe" North American first-day box office set annual record

This week "Spider-Man: Across the Universe" was officially released. The box office in North America exceeded US$51.7 million on its first day, surpassing the US$48.1 million of "Guardians of the Galaxy Vol. 3", which previously topped the list, and set a new record this year. The highest first-day record of the year in the North American box office market.

This first-day result ranks third among animated films in North American film history, behind "The Incredibles 2" with $71.2 million and "Finding Nemo 2" with $54.7 million.

The series chose Miles Morales, the little black spider, to succeed the classic Peter Parker and become the second Spider-Man in the Ultimate Universe, and opened the door to the Spider-Man universe. The movie scored 9.1 on Douban, and it still maintains a high score of 8.9.

# Welcome to follow Ai Faner’s official WeChat public account: Ai Faner (WeChat ID: ifanr). More exciting content will be provided to you as soon as possible.

The above is the detailed content of Morning Post | Apple may say goodbye to "Hey Siri"/Microsoft terminates support for Cortana/Ted Jiang says AI has no intelligence. For more information, please follow other related articles on the PHP Chinese website!