The Generative AI Revolution
As generative AIs are changing the world of creativity and content production, we are facing a true technological revolution.
🇮🇹 The italian version of this article is available here
Let’s talk about generative artificial intelligence, if you have also followed past issues of the newsletter you will not have been surprised by the overbearing prominence of these tools that I wrote about in my newsletter still last year.
It can be said that of the 3 heralded revolutions in digital, generative AI is the most successful, while web3 is holding back because of the collapse of the crypto ecosystem and adoption difficulties and the metaverse is virtually and completely depopulated of users and transactions.
The Generative AI Revolution
Generative Artificial Intelligence is a form of artificial intelligence that can create original content, such as text, music, and images, without the need for explicit input from a human being. Generative AI can be used in many fields, from automatic news writing to the creation of new musical melodies, and can offer many opportunities for innovation and creativity.
This technology has risen to the headlines for the resounding success and incredible performace of Generative AI tools such as DALL-E, Midjourney, GPT3, and in recent days ChatGPT and Lensa.ai
The issues of Generative AI
Enthusiasm for these technologies is high however, the use of Generative AI also presents some dangers and risks. First, there is the risk that Generative AI can be used to create false or misleading content, such as fake news or distorted data. In addition, the overuse of Generative AI could lead to the loss of jobs in the creative sector, such as writers, musicians, and artists, because of its ability to replace humans in some activities.
There is also a strong controversy in that the machine learning mechanisms used by AIs exploit millions of works of human activity either in the form of text, music or images to process the results, many artists have begun to argue that there should be some form of compensation for the uses of these works for creative purposes by an AI.
In addition, the use of Generative AI could have a negative impact on human creativity. While Generative AI can be used to generate new and original ideas, there is a risk that humans will become dependent on it and lose their ability to think creatively. In addition, overuse of Generative AI could lead to standardization of creative content, reducing diversity and variety.
The main types of Generative AI
In terms of solutions and companies in this area, there are several companies that offer Generative AI for creating any type of content and with a wide variety of input types including:
Text to Text — Generate a text from a simple command (Examples of companies in this area: OpenAI (GPT3 or ChatGPT), Jasper, Copy.ai, Writesonic.
Text to Image — Generate an image with a simple command (Examples of companies in this area: Crayon, Stability.ai, GauGan2, WonderAI, Artssy, Jasper)
Text to Video — Generate a video from text (Examples of companies in this area: Fliki, Synthesia, Runway, MetaAI)
Text to Audio — Generate an audio from text (Examples of companies in this area: Resemble.ai, Wellsaid.ai, Play.ht, ChatGPT, Murf.ai)
Text to Motion — Generate an animation from a command (Examples of companies in this area: Tree Ind.)
Image to Text — Generate a text from an image (Examples of companies in this area: Neural.love, GPT3 Demo)
Audio to Text — Generate a text from an audio command (Examples of companies in this area: Descrypt, Assembly.ai, Whisper)
Audio to Audio — Generate text from an audio command (Examples of companies in this area: VoiceMod, AudioLM)
Brain to Text — Generate text from a thought, and yes. (Examples of companies in this area: Neuralink, MetaAI)
Visuals to Ad — Identify objects in videos and link them to online ads, the future of advertising practically, imagine if Netflix decided to adopt it. (Examples of companies in this area: AI Lens)
Text to Code — Generate code from a simple command (Examples of companies in this area: Github Copilot, Replit)
Text to 3D — Generate a 3D object from text (Examples of companies in this area: GET3D, Dreamfusion)
Text to NFT — Generate from a command directly an NFT (Examples of companies in this area: NightCafé, Chaintext, Sketchar)
In this Linkedin post of mine you will find a downloadable PDF carousel that lists them all and gives a picture of the state of the industry, for example, it is interesting to see how much Venture Capital over the past few months is deciding to invest in similar solutions.
ChatGPT
What I would like to tell you about today is the latest from OpenAI, however, going deeper than the trivial wow effect hype bounced around on social these days.
ChatGPT is a technology developed by OpenAI that uses deep machine learning to simulate human conversations in real time. ChatGPT uses a pre-trained language model on a huge corpus of text to generate answers to questions posed by the user in a consistent and natural way.
When using ChatGPT, the user can start a conversation by typing a question or phrase into the input window. ChatGPT then uses its language model to generate a consistent and natural response, which is displayed in the output window. The user can then continue the conversation by responding to the response generated by ChatGPT, creating a kind of multi-way dialogue.
In addition, ChatGPT can be customized to fit specific contexts or topics. For example, ChatGPT can be trained on a body of text on a particular topic, such as art history, to generate more accurate and specific responses on that topic. In addition, ChatGPT can be used to generate original content, such as stories or articles, using its language model to create coherent and natural stories.
In conclusion, ChatGPT is a deep machine learning technology that allows human conversations to be simulated in real time. Using a pre-trained language model, ChatGPT can generate consistent and natural responses to questions posed by the user, creating a multi-way dialogue. In addition, ChatGPT can be customized to fit specific contexts or topics, and can be used to generate original content.
What limitations does Chat GPT have?
Let’s try to demystify ChatGPT, OpenAI’s solution is a GAN (Generative Adversarial Networks) application that beyond the hype generated on social media has some limitations, I would also like to mention that this is only a beta version in free testing, likely that the final version in addition to being better will be paid with a credit system.
For those unfamiliar with GANs, they are a type of neural network that uses two competing networks-a generator and a discriminator-to create realistic-looking output. The generator creates fake outputs and the discriminator tries to distinguish between the fake outputs and real-world data. Through this back-and-forth process, the GAN is able to produce output that is indistinguishable from real data.
ChatGPT takes this concept and applies it to text-based conversation. It uses GAN to generate responses to input text, allowing it to engage in conversations with a distinctly human touch.
But here’s the catch: ChatGPT’s responses touch reality only at a tangent. While they may sound convincing, they are ultimately fictitious creations of the GAN.
This might seem like a disadvantage, but it actually makes ChatGPT incredibly useful for all things creative and fictional. Because it is not bound by the constraints of reality, ChatGPT can engage in completely imaginary conversations and provide creative, ready-to-use responses.
Generative AI will be a heralded success
So why use ChatGPT if its answers only touch reality at a tangent? Because sometimes, it is exactly this kind of creative and imaginative thinking that we need to solve complex problems and generate new ideas. ChatGPT allows us to explore possibilities beyond the limits of our everyday reality that can be incredibly powerful.
ChatGPT went viral and reached 1 million users in just 5 days. Just to make a simple comparison, here is how long it took other products to reach 1 million users:
Netflix — 41 months
Twitter — 24 months
Facebook — 10 months
Instagram — 2.5 months
ChatGPT can among other things also write code snippets, but for now better to test them before using them, it is still a demo. You can try it out here:
👉 https://chat.openai.com/chat
Conclusion: Generative AI, risk or resource for creativity?
Let’s ask ChatGPT directly:
