Seven predictions: How will generative AI reimagine the world?

Author: Lu Aifang Science and Technology Capital

Since OpenAI launched ChatGPT on November 30 last year, the wave of AI has swept the technology world and even the entire society. Being able to write articles, draw pictures, write code, and pass the doctor's lawyer practice exam, people are both curious and excited and worried in the face of such powerful generative AI.

Generative AI is seen as a technological revolution comparable to (car) engines and personal computers, leading to another leap in human productivity. So how exactly will AI reshape the way we work and live? How will it reshape industry and society? How will the world change as a result? Science and technology capital summarizes seven core views and shares them with you. (Note: The cover image was generated by Midjourney)

1. Search is no longer the first entry

I've been experimenting with chatGPT since its launch, and in just a few months, behavior patterns and mentalities have changed. Now, the first thing I think of asking ChatGPT, especially in an area I am not familiar with, can help me build a framework first, and then do a targeted search.

After Microsoft integrated ChatGPT into Bing, Bing's traffic increased significantly, while Google's traffic declined. The trend is already clear that the search box will no longer be the first entry, but it will not disappear either, but will gradually become a vassal of chatbots like ChatGPT.

2. AIGC will become the standard for Internet products

After Microsoft Microsoft 365 is connected to ChatGPT, China's largest office platform DingTalk is also connected to Alibaba's Qianwen model, whether it is writing articles, doing PPT, or holding meetings, and answering emails, AI can help you.

With the development of OpenAI, an AIGC industry has been formed in the United States, and many startups provide various content generation services based on ChatGPT capabilities, such as Jasper focusing on ad copywriting, Waymark focusing on video generation, Inworld focusing on game character creation, and so on. If OpenAI is compared to Apple, they are apps in Apple's App Store.

Although AIGC has only just started, it has already developed a mature business model. OpenAI adopts a MAAS (Model as a Service) model, charging startups API interface fees based on usage, while startups charge users monthly or annual subscription fees. Compared with many apps in the early days of the mobile Internet that could not find a profit model and had to rely on venture capital to survive, this round of AIGC's startups are much luckier and the speed of marketization should be faster.

OpenAI has launched a ChatGPT plug-in for third-party applications, and many AIGC native applications will also force existing Internet products on the market to launch AIGC, so that AIGC can be rapidly popularized and become the standard configuration of Internet products. In the future, if you do not have AIGC capabilities, will you still use them?

Recommended Articles:

AIGC will become more popular than we think

What does the "iPhone moment of AI" really mean?

3. The Internet will be reconstructed

The emergence of ChatGPT, a big language model, will deconstruct the existing Internet. First, the user interface will be getting simpler. Now there are many various apps, everyone looks for the function they want in the complex menu, and what you want to do in the future, just use natural language to communicate with the AI. For example, if you want to prepare a table of dishes to entertain guests, ask the AI directly, it will tell you which dishes to prepare, what ingredients you need to buy, and then help you place an order on the shopping website, and finally generate a picture of the family banquet to help you post on social networking sites.

Second, with the access of large models, the underlying architecture of the Internet will also undergo fundamental changes. Now the data between apps is not connected, and everyone builds high walls to encircle users. But in the AI era, users bypass the boundaries of apps through natural language and directly reach data. Apps will recede into the background or even disappear, and only a few super apps will survive.

In the face of the powerful reconstruction power of generative AI, Internet companies have to think hard about how to use AI to transform their products as soon as possible and strive for greater living space for the future.

Recommended Articles:

How do I redo it with a large model? DingTalk has already figured it out

What happens when ChatGPT opens WeChat?

4. Either more powerful or cheaper than AI

After the advent of ChatGPT, white-collar workers shivered and worried about losing their jobs. According to Goldman Sachs' analysis, offices and administration are the highest proportion replaced by AI, reaching 46%; Law followed, at 44 per cent. In view of the fact that many AI applications emerging in the market focus on advertising and marketing, copywriters and designers are also high-risk groups.

But after all, generative AI has just been launched, often makes mistakes, and the understanding of human intentions is not accurate enough, and it cannot completely replace humans for the time being, and is more used as auxiliary tools. Many artists and designers use AI to quickly generate sketches with a large amount of inspiration, which greatly speeds up the efficiency of creation.

But AI learning evolves quickly, and sooner or later those simple, repetitive tasks will one day be completely replaced by AI. Workers must hurry up to learn, you are either more powerful than AI, after all, humans have the creativity and imagination that AI does not have yet. You are either cheaper than AI, after all, AI also has to pay, especially in the early stages, because computing power is expensive, AI services are not cheap.

In addition, there is no need to be too pessimistic, just like any technological revolution in human society in the past, AI will steal some jobs while creating some new occupations. For example, the information technology revolution of the 1990s gave rise to new professions such as web designers and software developers, and now prompt word engineers and AI trainers have become very sought-after.

5. Generative AI will further exacerbate the gap between rich and poor

Generative AI will bring a leap forward in human productivity, but it will also be very expensive in itself. The initial investment of training a large model such as chatGPT is nearly one billion US dollars, and the training costs 150-2 million US dollars, and the daily electricity cost is nearly 50,000 US dollars.

OpenAI was founded as a nonprofit in 2015 and announced in 2019 that it transitioned from "non-profit" to "limited profit." OpenAI now charges API interface fees to companies using its technology while charging users subscription fees.

There is no free AI in the world, and the free trial is likely to not respond, because there are too many people trying it. For example, in order to give this article a cover image, I sent a command on Midjourney, and the robot replied to me unceremoniously: I am very busy now, and I don't have time to respond to your free needs. I had to buy a membership for as little as $8 a month, pay for a year, and then the robot started working for me, is it like a human?

At present, the charging standard of AIGC application is basically tens of dollars per month, which may be affordable to everyone in developed countries in Europe and the United States, but it is not a small amount for the poor, especially in the backward regions of Asia, Africa and Latin America. According to the World Bank, the net per capita income of low-income countries in 2020 was only $650, as low as $350 in the Central African Republic. Such an income is difficult to even survive, let alone buy AI services.

This will create new inequalities, and companies that can adopt AI will gain higher productivity and more competitive advantages, crushing those that cannot afford AI. For example, I interviewed a virtual fashion company that cut the time it took to design a garment in half after using chatGPT and Midjourney. AI will further widen the gap between the rich and the poor, developed and backward countries, leading to a further widening of the gap between the rich and the poor.

Goldman Sachs believes that when about half of U.S. companies adopt generative AI, labor productivity growth in the U.S. could increase by about 1.5 percentage points a year, comparable to the invention of the (automobile) engine and personal computer. And for more lagging countries, when will generative AI be available?

In the past, technology companies have been shouting the slogan of technology inclusion. Indeed, whether it is a computer or a mobile phone, manufacturers will launch products with different configurations and different prices according to the consumption power of different regions, so that technology can benefit the world. And for generative AI, how will technology giants such as Microsoft and Google use it to benefit more people?

Recommended Articles:

Can AI Save a Declining Global Economy?

AIGC tide is raging, who profits first?

6. Generative AI will form the polarity between China and the United States

After OpenAI launched chatGPT, Chinese companies followed suit, competing to launch their own large models and chat applications, including Baidu Wenxin, Alibaba Tongyi Qianwen, Huawei Pangu, and SenseNova.

China does not lack large models, and the scale of training parameters is not small. For example, the training parameters of Baidu Wenxin model reached 260 billion, and the Ali Tongyi model even reached 10 trillion.

But after all, the big language model began in the United States, in 2017 Google first proposed the NLP model Transformer, in June 2018 OpenAI released the GPT large model, in October Google launched the Bert large model, in 2019, meta, Microsoft also launched their own large language model, and China's large model mainly in 2021-2022.

In addition to being one step late, the quantity and quality of Chinese data are not as good as English data, which will also affect the training of large models. In addition, in August 2022, the US government restricted the export of some high-end GPUs to China, which affected the training speed of the model to a certain extent.

So overall, Chinese companies lag 2-3 years behind the United States in terms of big language models.

But looking at the world, only China can wrestle with the United States. According to the statistics of the domestic OpenBMB open source community, at present, Chinese enterprises or institutions contribute 1/3 of the global more than 100 billion parameter model, the United States contributes 1/2, and other countries or organizations in the world contribute the remaining 1/6. As of October 2022, among the top ten organizations with a large number of models and the number of parameters, China and the United States each occupy 4/6 seats.

China has cloud computing companies such as Alibaba Cloud, Tencent Cloud, and HUAWEI CLOUD to provide infrastructure support for large models. China's industrial digitalization level is also increasing, and the scale of China's digital economy reached 45.5 trillion yuan in 2021, accounting for 39.8% of GDP. China also has a developed Internet industry and rich AI application scenarios, all of which provide favorable conditions for China to develop its own large model.

Recommended Articles:

The power of ChatGPT is inseparable from the top-level design of the US government

Is it possible to make a Chinese version of ChatGPT?

7. Generative AI will lead to the metaverse

Last year, the metaverse was a big hit, and this year it seems to have receded. But in fact, the emergence of AI is precisely to prepare for the birth of the metaverse.

The metaverse is a mapping of the real world in the virtual world, and the training process of generative AI itself is a mapping of the human brain, allowing machines to become human-like responses and thinking abilities.

With generative AI, pictures, pictures, videos and even 3D models can be generated with text, people can easily create their own digital avatars, digital spaces and various virtual items in the virtual world, and the threshold for mapping the real world to the virtual world is greatly reduced.

If combined with blockchain technology, these digital contents can be guaranteed and confirmed, and become digital assets for people in the virtual world. Generative AI also has the ability to write programs that help us create smart contracts that automatically flow and trade digital assets in the virtual world.

If generative AI is a revolution in productivity, blockchain is a revolution in production relations, AR\VR is a revolution in display terminals, these technologies are still growing independently on their own roads, but one day they will be stormy, and then the meta-universe will really come.

Recommended Articles:

OpenAI+Roblox may be the perfect combination of the metaverse

ChatGPT, Web3, smart society and metaverse

What other impact do you think generative AI will have? Welcome to leave a message in the comment area to discuss. Finally, remember to pay attention to technology capital, we grow with AI.