By now, you probably tried out ChatGPT, generated images on Midjourney or human-like videos on Synthesia. These tools demonstrate the potentials of generative artificial intelligence (GenAI) and provide a glimpse into what’s coming.
The true power of GenAI lies in its capabilities to have human-like conversations and generate human-like outputs. It’s still early, but as the technology advances, GenAI has the potential to be the next transformational technology to change not only how we do everything, but what we do. I have described below 3 reasons why I am excited about GenAI and its potentials to disrupt and create new value, likely at the scale of mobile and the internet.
1. Technology-Problem-Fit
Historically, transformational technologies introduced new capabilities and solutions to unsolved problems and found a technology-problem-fit. As a result, they scaled exponentially while disrupting the status quo and creating massive value for people and businesses. In the case of GenAI, what’s the technology-problem-fit? Let’s unpack that.
PROBLEM
Smartphones made it easier for people to get what they want, and softwares made it easier for companies to run their businesses. However, there are many unsolved problems for consumers and businesses that are too complex and require human-level intelligence hence done by humans, for now. To put into perspective, an average company with mostly white-collar workers spend x100 more in salary than software.
So the unsolved problems are the tasks, roles, and even industries that are done by humans - that’s a massive opportunity for GenAI. Which problems will be productized into a defensible business model? That’s a separate question that I am exploring as the focus of my next post. So stay tuned!
TECHNOLOGY
AI has been around for quite some time, but for most companies, it has been challenging to reap the benefits due to high barriers such as difficulty in finding and hiring AI talent, lengthy and costly to train and develop effective AI models. Consequently, only a small percentage of companies were able to leverage AI for high-value use cases. But, that’s about to change.
We saw two significant leaps in recent years. First, AI and large language models (LLM) have surpassed human capabilities in certain areas, most notably in reading comprehension and natural language understanding (see chart below). Second, tools like ChatGPT have made it very easy for anyone to use and benefit from GenAI through simple user interface.
Sam Altman, the founder of OpenAI, said powerful AI models will be the next tech platforms, paving the way for an explosion of new companies, similar to Apple and Android platforms did for mobile apps like Uber and Instagram. He predicts the new value creation will be at the scale of the App Stores.
Today, we already see use cases of GenAI demonstrating its technology-problem-fit. For example, Seek AI enables companies to ask and get answers from their data without the need of a data scientist and touching any raw data. Runway ML is super impressive and enables anyone to create and edit videos using commands in natural language - see the video below to see the GenAI magic in action. In the coming years, the capabilities of GenAI will likely compound exponentially and will bridge the gap between the technology and more complex problems.
2. Speed of Adoption
The launch of Dall-E and ChatGPT were the inflection point for GenAI, similar to the launch of iPhone was for mobile. In just 2 months, ChatGPT reached 100Mn users, a feat that took the internet ~10 years and mobile technology ~6 years. Even the most viral apps like TikTok took 9 months and Instagram took 30 months to reach 100Mn users. But these are not apple to apple comparisons. Let me unpack 2 reasons why GenAI adoption will be different and faster than anything we have seen before.
NO DEPENCENCY
Most transformational technologies, including their viral applications, rely on underlying hardware and infrastructure. For example, the internet depended on personal computers, and mobile apps needed smartphones. But today, almost everyone has a smartphone or computer hence can use ChatGPT.
GenAI use cases have no dependencies and barriers, except for compute power and cost. It's like the freeway is already built and all roadblocks have been removed. Also, ChatGPT is just one use case, and as hundreds of killer products and use cases emerge, GenAI adoption is set to skyrocket.
LANGUAGE INTERACTION
To reach mass adoption, no matter how great the technology is, an average person should effortlessly understand and use it. The iPhone, for example, is one of the best innovations in product design and user experience. Still, people had to learn how to interact with it, such as typing on a small screen and figuring out how to install an app. But with GenAI, the interaction is through natural language, as if talking to a human. You probably spent more time thinking about what to write in your first prompt in ChatGPT than figuring out how to use it. There will be no new learning curve for people to use GenAI, which will further accelerate adoption.
Today, those who speak the computer language harness the power of technology. But computers are learning to speak our language, so that power will be democratized. For me, the most fascinating aspect of GenAI is that it brings together the most powerful technology to solve the most complex problems, yet the user interaction is beautifully simple and intuitively natural.
3. Scale of Impact
Chamath Palihapitiya, founder and CEO of Social Capital, has identified two important trends for the next decade. First, the marginal cost of energy generation will drop to zero with the proliferation of solar power. Second, the marginal cost of compute is going to zero as ASICs, FPGAs, and GPUs will double compute capacity every 18-24 months. These trends will remove the computational capacity and cost constraints for GenAI, leading to an exponential acceleration in its advancement and adoption.
We will likely see a Cambrian explosion of startups across both axes, horizontal and vertical, bringing GenAI applications to solve complex problems. While it is highly uncertain what the aggregate net impact of this transformational technology will be on jobs, industries, economies, value creation vs destruction, and value distribution across consumer surplus vs businesses, it is certain that GenAI will lead to an unprecedented productivity increase by a factor of multiple digits. Let’s unpack that with an example.
According to Emad Mostaque, founder and CEO of Stability AI, we are ~5 years away from generating movies in real time without the need to write scripts. If this becomes the reality, we could see personalized fan versions of the ending to the Game of Thrones. Personally, I would be excited to create the fourth series of Richard Linklater’s Before trilogy. While this may seem far-fetched, ten years ago, Dall-E and ChatGPT would have seemed just as improbable. YouTube and TikTok have democratized content creation while capturing a significant share of watch time. Now, imagine a scenario where everyone can create content without turning on a camera.
From the example above, just imagine the impact on the cost of movie production. Imagine the impact on Hollywood, streaming services, and the entire entertainment industry. Imagine what kind of conversations Bob Iger at Disney, Reed Hastings at Netflix, and Shantanu Narayen at Adobe are having in their 10-year strategy refresh sessions. And imagine the impact across all other industries. I will end this post with those imagination.
Thanks for reading! To receive future posts, subscribe here: