How does AI redefine creative tools and media?

▎ Editor 's note:

Some people are proud of TRUMP coins. Someone is frustrated. Aside from FOMO's emotions, Meme is just the entrance. AI is the future of the spring of the chain. Grasp the most critical trend. The world is in my hands.

Today, I will share a " Neural Media" from the encrypted VC@BaincapCrypto. The author @NataLie thinks about the impact of generating artificial intelligence and cryptocurrencies on creative production, hoping to help everyone get some inspiration when looking for the next opportunity.

🎯 The main bright spots

1️⃣ Genesis artificial intelligence is leading to profound changes in the field of creative production. Its impact can be compared with the "NAPSTER Moment" with the cost of distribution of media in the Internet era to zero:

• The core of this change is that the cost of creative production is reduced to zero, which directly impacts the core of human creativity.

• Under the new paradigm, humans should start from focusing on the final output to the system and process, that is, teach neural network thinking at the programming level.

2️⃣ Through programming, we can create a unique "software brain" and generate unique ideas and works. Application scenarios include:

• Media based on Agent: Model simulation human partners, interacting through text dialogue, can perform financial transactions and other operations.

• Real -time game engine: Model simulation game engine, generate game frames according to user actions to achieve real -time rendering.

• Multi -universe generator: Model generates an unlimited change version, expands the original idea of ​​users, and explores the possibility space.

3️⃣ A trend facing the future may be:

• Creative tools: Prompting is embedded in more interfaces to stimulate end user creativity. Most Prompts will be abstracted into control, but creative vision, accuracy, taste and skills will be more important.

• Media business model evolution: From corporate media to users to generate media, to machine generation media. In the future, the consumer media business model will generate media establishment around Agent (innovative scenes include chat robots such as Character.AI, interface generation, such as websim, users to produce currencies such as pump.fun, etc.)

• Intellectual property challenges: Machine learning enables procedures to "learn" the aesthetic style of human creators. The cost of creating production and aesthetic imitation to zero, the value and significance of intellectual property rights need to be reviewed.

4️⃣ The role of cryptocurrencies can play include:

• The market on the chain to generate the media to generate the media (such as the nearest DEFAI);

• As an intellectual property incentive layer;

• Media monetization and access control, such as Minting becomes a new business model; NFT can be used as a personal program and user -generated software infrastructure;

• As a economic coordination layer between human -computer social, explore a new paradigm for community operations and Agent interaction.

All in all, this is an article that may be difficult to read but is worth thinking about. AI makes human creativity more reflected in the design of systems and processes, and cryptocurrencies provide new economic and society for this change. Coordination mechanism. In the next media era, what new opportunities and new trends can be created by the two can wait and see.

► ► ► ►

是 "All media are an extension of a certain ability of human beings -psychological or physiological." ~ Marshall Mc Luhan

Most of the time in 2024, I spent a lot of time trying to understand what we are now calling " generating artificial intelligence " and its impact on me and the entire society. I am deeply attracted by artificial intelligence as creative tools, and uses these new products in the workflow, especially in creative writing and music creation.

However, as an encrypted investor focusing on consumer media and applications for users, artificial intelligence is more and more like a blind spot for me. When we talk about the most successful consumer media companies in the Internet era, we will not discuss them from the perspective of technology islands, because they are not built like this, just like Facebook's success is inseparable from technological innovation, but we will not take Facebook Instead, it is purely regarded as a "mobile application" or "artificial intelligence application". On the contrary, we realize that it is the convergence of many different innovations that make applications like Facebook possible.

In this context, this article aims to integrate and improve my personal discovery and insights that I have explored artificial intelligence in the past year. I share these contents, hoping to resonate or help others (especially my colleagues in encrypted enthusiasts).

Part.1, another "napster moment"

Today, discussions around artificial intelligence generation media are mainly concentrated in: (1) the ethical issues of model training and data capture, (2) whether "artificial intelligence art" is the real art, and (3) deep -falsified anti -utopian prospects Essence These discussions are very interesting and worth listening to it. However, I think they saw that the trees were not seen in some important aspects.

I found that the most useful framework for understanding the rise of genetic artificial intelligence is that it is regarded as an intellectual property rights in another "NAPSER time" (NAPSTER is the first point -to -point music sharing service that is widely used. People, especially the way college students use the Internet), but this time is the time of production, not dividing time.

The rise of the Internet and the process of subsequent media distribution costs to zero are a moment of "from nothing to". The suddenness of this transformation was wonderful in the documentary "How to Get Freedom of Music". The film tells the story of a CD factory worker and a group of young hackers pushing the entire music industry to collapse overnight.

Before the emergence of NAPSTER and a broader digital file sharing, the entire corporate media, industrial complex (and the livelihood of artists) depend on the technical reality of expensive, high friction and concentrated technology. In just a few years after its launch, large record companies have transformed from a record sales volume to begging for the federal government to save them through legal intervention. The industry faces an extremely difficult reality: the economic system that supports its business has undergone fundamental and irreversible changes, and the era of buying music has ended .

Today, I think that generating artificial intelligence has brought us a more difficult reality. The impact of creative production costs to zero is more difficult to cope in many aspects, because it directly touches the core of many people think that it makes us human: Our creativity . This kind of existence fear does not change the following facts: media generation (especially "style migration" or aesthetic imitation) is free, including all media types (text, images, video, audio, software) that we are concerned about now - - This is another moment of "from nothing to".

However, the most important difference between today and the early 2000s is that in the struggle between Napster and media companies, the government stood on the side of the company and eventually shared the documents as "piracy" behavior. (This is why we often call corporate media/intellectual property rights "legal media"). This decision, as well as Steve Jobs launched the iPod, promoted it later to become iTunes products, and eventually developed into a "streaming media", which made the industry completely collapse. Unfortunately, I think those creators who are expected to involve and take action here are at best to comfort themselves, and the worst is to deceive themselves .

I think we may find that the intellectual property system is mainly to protect the company and its legal media, and no one will come to save us . Traditional media companies have learned painful lessons last time, so they took the initiative to perform licensed transactions with artificial intelligence companies and were compensated to a certain extent. New media companies are also using users to generate content model training shared on their platforms, even if they claim that they have not done so. However, independent creators are largely left behind.

Part.2, calculation: the medium of our time

It is easy to understand why many creators think that generating artificial intelligence has weakened their abilities. I think this concern is largely reasonable. However, I also think that there is an opportunity to think that calculation is developing in a new way, which not only requires us to regard it as a medium of exchange, but also call on us as a creative medium .

For those who have created video games or generate art, the concept of "calculation as a creative medium" is not fresh. However, many people have not really realized this today. Software is the first digital native media category. Most people mainly understand it from the perspective of "service", "practicality" and "optimization", not necessarily from the perspective of creative expression. Now, generating artificial intelligence is promoting this view in a very direct way. It has reduced the production cost of almost all other media to zero . This seems to trigger a sexual problem: "So, where is human creativity? Where is the value of manual skills?"

My answer may not be surprising: " It is reflected at the programmable level ." Before we discuss what I mean, we need to understand a few important technical concepts.

10 2.1 neural network 101 (suitable for beginners)

Training is a process. In essence, it is to "teach" the model to complete the task by providing a large number of examples that complete a task, then let it find the mode, predict the new input, and correct it when making mistakes. In conceptual point of view, this is similar to our learning painting: starting through imitation shape until we can create original works, and at the same time use the feedback from our companions and teachers to continuously improve our skills. Of course, there is a key difference: for example, text generation models do not learn to write like you and me, but learn to simulate writing with high accuracy. This is why I increasingly agree with "Simulators" rather than "agents" is one of the many reasons that are more suitable for neural networks.

Potent space , or I prefer to call it "high -dimensional possibility space", which is a representation space in the neural network. The content learned during the training process is presented here in compression. For example, this is similar to the "internal world model" built in the complex relationship between the model when learning and understanding the complicated relationships between detectable characteristics. Understanding the concept of potential space is the key to understanding neural networks as creative tools and media .

 Potential space visualization#1 -interpolation between known embeds 

 Potential space visualization#2 — Representation of multi -dimensional attribute & relationships in different embedded

Embeddings : Embeds can be regarded as a process of mapping the input into a specific point in the potential space. This is the process of translating prompts into a model "thinking language" in nature. In this way, we can understand the "prompt" as a way to explore and navigate the potential space of navigation model -this means that proficient in the prompt is to form intuition for the potential space shape of the model, so as to guide the model to generate specific, expected expectations Output .

One of the fun of playing neural networks is that their deep internal working principles are still a mystery for us. However, I think these basic concepts provide the necessary background for the consistent network as creative tools.

Part.3, neural network: a new innovation paradigm

One of the core points of the computer media is that it requires us to shift from the final output (songs, images, videos, texts) to more attention to the system and process . Specific cases of neural networks means that we need to think of it as a programmable media generating engine, rather than simply regarding it as a tool for a certain media. From this perspective, I discovered the answer to the question of the above "the value of human creativity and process": it exists in the design of the training process and model architecture -this is what I call "at the level of programming" .

XhairyMutantx is a work created by Holly HernDon and Mat Dryhurt -this model is strictly trained based on Holly -based photos. No matter what the input prompts are entered, it will generate photos inspired by her appearance.

If you regard the neural network as an attempt, that is, the abstraction of software -based human cognitive functions, so obvious, training and design models are equivalent to teaching it how to think.

You can imagine to send an instruction to all friends ("Tips"): "Recall a childhood memory." Each person's answer will obviously be different, because the content they generates will depend on their personal background and imagination (that is, "training" training " data"). After many prompts, you may also find that some friends can always generate more beautiful or creative answers, and may even show some specific personal style. So, if you can use all the human brains in the world to perform this exercise? If you can choose a particularly unique human brain, such as Picasso or Kane West?

This is essentially the creative super ability provided by the neural network -the ability to use other ideas as a creative tool . Here, I think it is not the specific output of a certain model that is really noticeable, but a "software brain" that has the opportunity to create a "software brain" creatively , which can produce unique ideas and unique works.

Arcade.ai is a "from prompt to product" market that allows users to design their own jewelry products. They specially adjusted a model to generate jewelry images that generate high -fidelity authenticity, and these images only use materials that end users can be used for manufacturing.

Further exploring the view that " systems are more important than output" and another remarkable feature of interacting with neural networks is to participate in a continuous feedback loop of prompts and response -I hear some people compared it as reading and writing than reading and writing. Feedback cycle. I personally noticed that I rarely ended an interaction after sending a prompt to the model. After receiving an output. Almost every interaction with the model will bring me into this interactive feedback cycle, so that I constantly iterate, reflection and exploration . This seems subtle, but it is a key to understanding the type of media generation media:

理 3.1 Agent -based media

I briefly mentioned this concept in a previous article, and the core idea is very simple -here, the model simulates the role of a human partner, interacts with us through text dialogue, and at the same time, it can also understand and in other forms The media responded. We can also see some models here that can represent other people or it takes action (such as performing financial transactions). Typical cases include chat robots, AI partners, NPCs (non -player characters) in the game, or any other anthropomorphic user experience. For example, Andy Ayrey's creative experiment "Infinite Backrooms" is a particularly interesting case by setting up multiple Claude instances for unsmodied intervention.

引3.2 real -time game engine

Here, the model simulates a game engine (or more specifically, a game state conversion function). By receiving user actions in the game as a prompt, the next frame response output in the game is generated . If the speed is fast enough, this experience should be similar to navigation in a virtual world, and this world will render in real time according to your actions. This is the ultimate expression of immersive and interactive media .

Doom game frame is generated by Gamengen . Gamengen is a game engine that is completely driven by a neuro model, as described in Google's "diffusion model is a real -time game engine" paper.

宇 3.3 multi -universe generator

In this scenario, the model plays the role of creative "deity". By generating unlimited changes to help us expand the original idea , each version can be further explored and controlled. This enables us to explore the possibility space around it from any ideas or concepts. For example, AI DUNGEON (a text -based "Choosing your own adventure" game) is an excellent case in this area.

LOOM 's user interface view, LOOM is a tree -based writing interface, suitable for language models like Chat GPT, provided by @Repliguate.

空3.4 potential space as a creative tool

I am increasingly believed that the concept of "exploration possibility space" is the core of understanding neural networks as creative tools and media . In the process of using tools such as Midjourney, Suno, Websim, Claude and other tools, I noticed that most of my operating processes can be attributed to the following modes:

Tip (Prompt) → generate a variant of specific output → use variants as a prompt for new output → regenerate into specific variants → such cycle and reciprocity ...

For example, when using AI -driven music generation tool Suno, I usually provide a 60 -second personal singing example and some written lyrics as prompts. Then, I will use the COVER function to generate an output, then regenerate the more than 10 variants of the output, and choose the part I like from these variants as the input of further prompts.

In essence, I am in the possibility space of my personal example in the potential space of the exploration model -discovering the variants based on my original work, these variants may not be able to think of it by myself, or to complete it in a reasonable time in a reasonable time. of. I think this method has unlocked an unprecedented fast prototype design and creative test process, and will give birth to "100 times the creator", similar to the "100 times the engineer of AI assistance" discussed in the software field.

I clearly realize that potential space is a creative tool. The use of artificial intelligence for creative production is not only in the model with powerful training functions, but also to design the interface that can empower users to explore and control these broad potential possibilities space with higher accuracy and fine particle size.

Part.4, consumer behavior and cultural impact

I have the following three predictions about how this technology changes consumer behavior and what new business opportunities it has created:

作4.1 will become creative tools

Prompting -Whether it is text, image, or other forms -this way of interaction is gradually embedded in more and more interfaces and experiences, and the creativity of end users is introduced into the areas that have never been involved in the past. Scott Belsky pointed out that "the early" early '' based on the PROMPT ' -based generated text to the image weakened the creativity, and the "Controls" era released human creativity in an unimaginable way. Skills will be more important than ever. "I agree that most Prompts will eventually be abstracted into" control "(components with user interfaces), so that users operate unconsciously. But more importantly, I think this trend has fundamentally changed our way of thinking about interface design.

**体4.2 Enterprise Media → Users Generate Media → Machine Generate

Media**

The last major change in the media business model was from the emergence of the media from an enterprise to the media. Now it seems that the next main consumer media business model will be established around the popularity of machines to generate media . However, it is still unclear what the "winner" will look like. Will it be a general model like Midjourney? More specialized creative tools? Or is it based on social experience above these technologies? Or is it a third choice that is even more conspicuous?

In any case, if you are the founders or independent creators in the current consumer media field, you may need to make a strategy to think about how to use these tools to enhance your business and promote growth.

In addition, I think another area worthy of attention is: how to make the AI ​​-driven experience more social and multi -user collaboration . Taking my personal experience as an example, most AI applications today are very "anti -social" because you are mainly interacting with the model, not interacting with others . There may be many opportunities and design spaces in this field , such as building a human -centered collaborative creative experience, or creating a new way to make humans and robots realize more meaningful social interaction.

产4.3 The impact on intellectual property rights

Not only the cost of creative production is falling to zero, especially the cost of aesthetic imitation to zero. I can take a picture of a person wearing it and enter Midjourney as a prompt to design a sofa with the same style. I can also migrate similar styles of the person's voice, writing style, etc. Under this new paradigm, what is the value and significance of intellectual property rights?

I have not found the answer, but obviously, most of the previous assumptions and thinking models are no longer applicable.

Part.5, the role and summary of cryptocurrencies

If you read here -thank you for your patience!

I will explore the impact of these contents on cryptocurrencies in future articles, but now I can preview the directions I will follow:

• The opportunity to build a crypto company around new media

Explore the potential of the market to generate media intersection on the chain.

• Cryptive as an incentive layer of intellectual property rights

Beyond the attribution and traceability, and think about building an incentive mechanism and network around the media.

• Crypto as the monetization and access control layer of the media

Especially in the field of software generated by users, rethink the web architecture; use "minting" as a business model of small models; NFT as an infrastructure for generating software to generate software for personal programs and users.

• Cryptive as a social and economic coordination layer between people and machines

Support humans and AI in identification, funding, and solving various issues; explore the models of community ownership and operation.