All the fuss about generative AI: money, possibilities, and pitfalls

VCs are spat at generative AI companies, opening checks for very large funding rounds despite the current economic climate. For example, in the second half of 2022, in the US alone, Jasper raised his $125 million Series A round and Stability AI secured his $100 million A round.

To put this into perspective, Cooley’s third quarter venture financing report showed pre-money valuations across Series A deals fell. So it’s pretty remarkable to see so many series A rounds in generative AI companies.

And with ChatGPT integrated to power the Bing search engine, there’s Microsoft’s multi-year investment in OpenAI, rumored to be as high as $10 billion. Companies like Alphabet and Nvidia are also said to be considering new investments in generative AI.

Some of the reasons behind this momentum are explained in a recent A16Z blog post by Matt Bornstein, Guido Appenzeller, and Martin Casado. As they wrote, “Models such as Stable Diffusion and ChatGPT have set historic records for user growth, with several applications reaching $100 million in annual revenue less than a year after launch. A side-by-side comparison shows that AI models outperform humans by orders of magnitude on some tasks.” Simply put, generative AI can be big business.

Generative AI is gaining significant traction globally, especially in the APAC region. In fact, Asia-Pacific is projected to grow even faster than the United States, at a CAGR of over 35% from 2022 to 2028. The market is driven by significant government initiatives and widespread adoption of AI-based applications. As we are held back more by imagination than by technology, this will further increase the demand for innovation and new products around the world.

But it’s not just engineers who are excited about generative AI. Everyday people are also fascinated. From school classrooms to local gyms to dinner parties, people are tinkering with his ChatGPT and creating images with Lensa AI to bring generative AI into the mainstream.

In fact, the paid app Lensa AI is gaining a lot of attention among consumers. Lensa AI made him $16.2 million in revenue in 2022, and in December 2022 alone he made $8 million. The app has been downloaded over 25 million times by him and has 1.1 million active users of his as of launch in December 2022. After all, ChatGPT has surpassed 1 million users in a week since its launch, and continues to attract users, rapidly gaining popularity.

So what does that mean and where is generative AI headed?

new creators

Content creation and entertainment are among the most common use cases for generative AI. People are excited to see text, audio, still images, and even video combined into AI models that they can share with friends, family, or the world at large. The prospect of building new kinds of avatars is particularly attractive if the Metaverse reaches its full potential. A new world is opening up.

Generative AI, for example, helps people without the relevant skills or expertise to use and create art. In the past, amateurish designs required skill with Unity 3D, Unreal Engine, Adobe Creator Tools, and more. Even the most limited editing required a certain kind of in-depth knowledge, training, or special equipment.

Today, we live in an age where everyone is a creator, but not everyone has the skill set to create great works of art. Generative AI will level the playing field and democratize art .

We don’t just use generative AI to build visual creations or create content like scripts and blog posts. There are also advances that allow multiple components to be put together, such as avatars that can be used in video and chat.

Why does this matter, other than just show it to your friends or show you another way to create a school project? Generative AI has the potential to be much bigger than that. there is. In fact, it helps people of all ages and all walks of life bring their imaginations into the real world.

But if you’re looking for a “big money” case, maybe a casting director is trying to consider the right actor for a movie role, or someone’s watching a YouTube video that could get millions of views. Think about it if you have an idea. You can now generate your “ideal” actor using avatar tools that can display emotions, change voice tone and expression (creating voice identities that don’t exist in this world). Or apply a licensed distinctive voice like Homer). before looking for human actors such as Simpson and James Earl Jones.

Use cases like this remove risk and save time by identifying the right traits and testing your vision before moving forward with a project. It’s much more accurate and efficient. They can see what works and what doesn’t. Ultimately, making the right decisions the first time can save you up to millions of dollars, depending on the scope of your project.

Drawbacks of generative AI

But like any new or evolving technology, there were problems. On the one hand, someone’s likeness or views can be adopted or misunderstood. Fortunately, sophisticated technical solutions are rapidly being developed to prevent unauthorized use. When someone creates an avatar to address a sensitive topic, they can apply example-based filtering algorithms to make the AI ​​system understand what the avatar would actually say or how they would answer. . These capabilities go well beyond traditional filtering algorithms that simply reject profanity.

The authoring tool can present examples of political issues and articles to the AI ​​system, and users can provide feedback (good or bad). Based on that feedback, the avatar decides what to say and what not to say (the approach is detailed in the research paper). User-defined content detection framework).

Similarly, OpenAI recently released a detection tool to distinguish between AI and human writing. The caveat is that it’s not yet 100% reliable. This is a very new area, so mistakes are still prevalent, but progress is being made. As complex issues such as copyright, intellectual property rights, and plagiarism continue to emerge, social consensus and ethics regarding the use of such technologies become increasingly important.

Artists, writers, and other creators say generative AI is a form of cheating that devalues ​​real art. A new lawsuit is going through court. A recent lawsuit by three of his working artists against Stability AI, image generator startup Midjourney, and online gallery DeviantArt alleges that AI image his generator “violates the rights of millions of artists in his 21st-century collages.” It claims to be nothing more than a tool.

Another lawsuit filed by Getty Images relates to copying images without a license and infringing intellectual property. The lawsuit highlights his two key issues with generative AI.

First, the definition of art. What constitutes art? Who decides what it is and how it needs to be created? Creating new generative AI is still a product of someone’s mind and experience. They are still expressions meant to provoke emotions and different reactions. Only the method changes. Artists do not draw with a brush, but with questions and orders.

Second: Who owns products created by generative AI? You can use images, audio, text, etc. that are properly licensed or quoted, and you can even work in the public domain. In addition, generative AI can draw from features such as voiceprints and facial expressions to build things that don’t and have never existed in the real world.

In the future, we expect to see far more license agreements that reflect changes in the world that incorporate generative AI and changes in laws related to intellectual property and copyright. As generative AI becomes more prevalent, this is inevitable.

Generative AI is in its infancy and there’s a lot we haven’t discovered yet. The potential is there to open up a whole new creative universe, but only if you ask the right questions and give the right orders in creation. As AI systems get smarter, so will the people who use them. With the right tools, the results will surprise us all.

Kim Tae-soo is CEO of Neosapiens.