Author |Rain Valley.
In the evening, Joan, a white-collar worker at a technology company, came home and was about to choose a good drama to enjoy a good bedtime time, when a new drama called "Joan Sucks" attracted her attention - the protagonist "Joan" in the play is the same as "Joan" from name to style. Not only that, everything that Joan experienced during the day was filmed in the play, and even suffered a "malicious interpretation".
Joan quickly found a lawyer and prepared to file a complaint. But she was surprised to find that neither the TV station nor the actors who played her in the show could sue them for infringement. This is because this is a ** show created entirely by AI, and the months required for manual production in the past have also been shortened to a few minutes by AI.
This was originally the story described in the first episode of the sixth season of the American drama "Black Mirror". But who would have thought that a few months after the show aired, the birth of the ** large model Sora made the story in the TV series accelerate into reality. When people experience the shock brought by ChatGPT in 2023, 2024 belongs to AI.
**Large model, open another spring of AI
In 2023, the GPT-4 language model released by OpenAI became the focus of the AI field and is considered a major breakthrough in the field of artificial intelligence. After a year, in the early morning of February 16, 2024, OpenAI officially released the Wensheng ** model SORA. Users only need to input text instructions, and SORA can quickly generate 60s ** content, while understanding and simulating the real world, and output ** that accurately reflects the user's prompts.
OpenAI positions SORA as a generative model for world simulators, laying the foundation for future development to understand and simulate real-world models. According to the official introduction, Sora currently has the ability to generate subjects and complex scenes that include multiple characters, specific types of movements, accurate details, and so on. At the same time, it is also possible to create multiple shots to simulate the complex camera movement effect, making the generated ** more vivid.
Screenshot of the ** generated by the SORA.
In addition, because OpenAI combines diffusion models with LLMs, SORA has the ability to understand and simulate the real world, so the ** generated by SORA can already show the real physical relationships between space and objects in space. At present, in addition to the Wensheng function, SORA also has the functions of picture generation, extension, splicing, editing, image generation and so on.
The release of SORA not only shocked enterprises and enthusiasts in the field of AI, but also immediately overshadowed Gemini 1., an iteration of the AI model released by Google on February 165 Pro in the limelight. The sadness of being "cut off" is probably only known to Google.
Originally, Google Gemini had worked hard to bypass the text link and directly cross over to the stage of understanding the world through sight and sound (although it was later confirmed that it was Google employees who helped Gemini generate responses through keyword prompts), thus avoiding the "battle" with GPT-4 at the same latitude. But OpenAI's approach of "not talking about martial arts" caught Gemini off guard.
Screenshot of Gemini's interactive presentation through visual and voice interaction**.
Of course, there are even more aggrieved: Gabor Cselle, an AI entrepreneur who has worked at Google and Twitter, uses Gemini 15. Analyzed the cherry blossom snow scene generated by SORA**. Didn't expect the gemini 15 is very pertinent to give the conclusion that "the visual effect of ** is amazing, but it is not shot in the real scene", giving people a sense of contrast that "even the opponent says it is good". However, Google also released on the 22nd that the average performance far exceeds llama, and claims to be the world's strongest open source model gemma, which can be regarded as barely gaining some face.
In addition to Google, film and game companies have also been on the verge lately. Zak Kukoff**, an early investor in San Francisco, USA, within 5 years, a team of less than 5 people will be able to make a film with more than $50 million at the box office using the Wensheng model.
Of course, there are also a lot of positives. Zhou Hongyi, chairman of 360 Company, directly posted a long post on the circle of friends, saying, "Sora's simulation of the physical world will at least have a huge impact on robot embodied intelligence (the ability to interact with the environment in real time through perception and interaction, according to the research note) and autonomous driving." It shows not only the ability of a first-class production, but also the ability of the large model to understand and simulate the real world, which will bring new results and breakthroughs. ”
In the first year of the big ** model started by SORA, there will definitely be more companies "abandoning the text and following the eyes". Whether it's good or bad, I'm afraid you can only know if you "see" it.
The agi version of "Game of Thrones".
In 2023, the "palace fighting drama" that broke out within OpenAI made many people who don't pay attention to the field of artificial intelligence know that outside of Japan, there is an "Ultraman" in the United States - 38-year-old OpenAI CEO Sam Altman (Sam Altman). And as OpenAI and Ultraman himself frequently gain attention in front of the public, a "Game of Thrones" that has attracted global attention and belongs to the AGI industry is also being staged.
In 2015, Altman co-founded OpenAI with Tesla CEO Elon Musk and others, and it started as a for-profit company run by a nonprofit board that prioritized mission over financial interests. Musk said at the time: "We want to have a Linux version of AI that is not controlled by any individual or company." ”
GPT-2 released in 2019 is an open-source model, but after that, OpenAI moved towards "closed source" under the leadership of Ultraman - its subsequent releases of GPT-3 and GPT-35, as well as GPT-4, are closed-source models.
Musk, one of the co-founders, has publicly expressed his dissatisfaction with OpenAI's board of directors in 2018, saying that OpenAI was founded as an open source company and is now a closed-source, profit-seeking company controlled by Microsoft. This is not what he wants at all. In 2023, Musk still said that the "closed-source" behavior of AI startup OpenAI is not a desirable approach.
Altman once explained: A long time ago, I realized that non-profit is not a long-term solution, and relying only on the funds raised by non-profit is far from enough for OpenAI's research, and commercialization is an inevitable choice.
Even to prove that OpenAI's non-profit nature still exists, he said at a conference: "No one in the company can be fully trusted." The board can fire me, and I think that's important. It's just that after the openai "palace fighting drama" happened, few people believe Ultraman's remarks anymore.
What is even more interesting is that Altman also said that the knowledge he understands AGI should be more than the sum of human scientific knowledge, and it should be creative and able to promote the development of basic science.
Ultraman's security risks brought about by AI commercialization and OpenAI's "palace fight" have further aroused global attention to AI ethics - in 2023, many countries and organizations began to formulate AI ethical standards to ensure the sustainable development of AI technology. These standards emphasize the importance of transparency, fairness, and explainability in AI systems, providing ethical guidance for future AI development.
In December 2023, the European Parliament, EU member states, and the European Commission reached an agreement to adopt the Artificial Intelligence Act, the world's first comprehensive regulation in the field of artificial intelligence.
With the release of SORA, the evolution time of AGI seems to have been shortened by 5-10 years, which also reminds many people of the words of Geoffrey Hinton, who was awarded the Turing Award as the "godfather of AI", regretting the development of artificial intelligence and worrying that artificial intelligence will bring serious harm to the world.
More than 20 years ago, Spielberg set up a "moral dilemma" based on humans and artificial intelligence (AI) in his film "Artificial Intelligence" - whether robots should be bound by moral norms when they have complex human emotions. In fact, this is also a discussion of the moral dilemmas that exist in the coexistence and interrelationship between artificial intelligence and human society.
Some industry insiders pointed out that in the current situation where the current governance framework and control measures have not kept up, enterprises alone may not be able to provide the AI security and trust required by society. However, for the AGI industry, Ultraman's choice may determine the direction of AI evolution in the future.
In an article published in the New York tech blog Intelligencer, author Elizabeth Wei called Altman "an imperialist disguised as a good guy." Paul Graham, co-founder of Y Combinator, a well-known American startup incubator, once gave a similar assessment: "You can throw him on an island full of cannibals and come back five years later and find him king." ”
In an interview with Time, a person familiar with OpenAI's board of directors said of Altman: "In many ways, Sam is really a good person......He cares about the mission, he cares about other people, he cares about humanity. But if you look at his behavior, you can also clearly see a pattern of using extreme methods to pursue power. ”
Judging from these evaluations, perhaps Ultraman only wants to be the king of the "Land of Light" and does not want to be a slave of AI. And for this agi version of "Game of Thrones", users who are currently spectators will also suffer the result of the "end of the show".
New energy vehicle racing, relying on AI to score points?
In terms of the combination of AI and daily life, new energy vehicles are undoubtedly one of the scenarios closest to reality, and they are also one of the most concerned tracks in recent years. Intelligent driving is not only a specific form of AI landing, but also the main direction for car companies to win industry competition. However, in an increasingly competitive market environment, car companies are facing increasing challenges.
On February 18, it was reported that Gaohe Automobile held an internal meeting and announced that it would suspend production for 6 months from now on. Although some insiders of Gaohe Automobile denied it in a timely manner, this news still triggered negative conjectures among some netizens about Gaohe.
In addition, Huawei's two models, the Zhijie S7 and the new M7, have been delayed in delivery, although Huawei has proposed the "car care subsidy plan", but there are still many car owners on social platforms to update the cumulative subsidy to complain.
Regarding the situation of Chinese car companies in 2024, He Xiaopeng, CEO of Xpeng Motors, said. "2024 will be the first year for Chinese auto brands to enter the 'sea of blood' competition, and a 'fight' in a fierce competition is inevitable. ”
Just after the Spring Festival holiday, BYD took the lead in launching the "** war" and lowered the prices of many of its models. Later, many car companies such as Wuling Motors, Changan Qiyuan, and Nezha Automobile followed up and announced price cuts.
Among them, BYD officially launched two new cars, Qin plus Glory Edition and Destroyer 05 Glory Edition, 7The ** with a minimum sale of 980,000 yuan makes the advertising slogan "electricity is lower than oil" particularly attractive; Wuling Starlight officially released the 150 advanced version of the Honor priced at 9980,000 yuan, so that the price of ultra-long pure electric range plug-in hybrid models officially entered the "90,000 level" era.
Many people**2024 will be the most volatile year in the new energy vehicle market, and the industry will completely shift from endurance racing to racing, but who would have thought that at the beginning of this year, car companies will be greeted by the "knockout race".
Previously, the "disappearance" of WM Motor has brought a new sense of crisis to new energy vehicle owners: the biggest risk faced by functions such as intelligent cockpit is that once the brand car company goes bankrupt and goes bankrupt, "intelligence reduction" is inevitable. The delay in delivery of mobile phones also reflects that there is still a big gap between its control over the industrial chain and the first chain compared with traditional car companies.
As an important commercial application of AI technology in the field of transportation, the implementation of autonomous driving will bring convenience to people's travel, improve traffic efficiency, and reduce traffic accidents. However, to date, the popularization of autonomous driving technology still faces many challenges, such as technical safety, laws and regulations, and social acceptance, and it is still difficult to become the core application technology of mainstream smart vehicles in the short term. However, as Zhou Hongyi said, the "birth" of SORA will have a huge impact on autonomous driving.
Of course, the beginning of 2024 is more uncomfortable with the Xiaomi car SU7 that has just been "launched". Lei Jun said at the press conference held on December 28 last year when explaining the performance of Xiaomi's first sedan SU7: "Don't shout 90,009, it's impossible!" He further said, "Anyone with this kind of performance and configuration will have to be more than 400,000!" So 140,009 doesn't need to talk about it anymore, you still have to respect technology! ”
However, this wave of new energy vehicle price cuts led by BYD may be a blow to Xiaomi, which is testing the upper limit of the psychology of potential car buyers. After all, compared with the "cost performance" of car companies that have been deeply involved in the automotive industry for decades, Xiaomi has no advantage at all. But if Xiaomi chooses to take the mid-to-high-end route, only selling a few hundred units a month, and then at the beginning of the year, it is rumored that it will "stop work" and take the high-end differentiated innovation route, I am afraid it will be a lesson from the previous "car".
The Chinese believe in a saying: "The car" must have a road before reaching the mountain. Market development has promoted the industry to accelerate the reshuffle, and the enterprises and brands that have experienced the survival of the fittest can drive the faster and better development of the industry. From this point of view, the "knockout" of the new energy vehicle industry at the beginning of the year is not necessarily a "bad" thing. And once the mobile phone gang with AI technology "survives", it is bound to "go to the big score" in the future.
40 years ago, James Cameron's "The Terminator" opened a new chapter in science fiction films, and also triggered the "whether the development of technology is lucky or disaster"**.
Today, while AI is showing great productivity, it is also entering all walks of life like a "terminator". And its development speed has far exceeded human imagination. We all have to admit that the era of AI is "in progress". As for whether it is "bad"? It's up to you to accept the "AI".