AI systems, typically expensive and complex, are mostly used by large tech firms. But imagine if your local pizza shop could use AI to predict daily sales. In a TED Talk , Andrew Ng envisions a future where AI is accessible to all, boosting profits and enriching society with just a few data points.
In the past, the ability to read and write was a privilege reserved for a select few. The majority of the population, mostly farmers and shepherds, didn't need these skills. The literate few, such as priests and monks, would read religious texts to the masses. As society progressed, we realized the immense benefits of widespread literacy, leading to a richer, more vibrant society.
Today, we stand at a similar juncture with artificial intelligence (AI). Currently, AI is like an exclusive language understood and used only by a select group of "high priests" - highly skilled AI engineers, primarily employed by big tech companies. But what if we could democratize AI, making it as commonplace as literacy? What if everyone, regardless of their profession or business size, could harness the power of AI?
Big tech companies have been the primary drivers of AI because developing AI systems is an expensive endeavor. It requires a team of skilled engineers and significant financial investment, often running into millions of dollars. These companies can afford such investments because they cater to hundreds of millions, or even billions, of users. They can create a "one-size-fits-all" AI system, like a web search engine or a product recommendation system, and apply it across their vast user base to generate substantial revenue.
But what about small businesses that don't have millions of users? Consider the local pizza store owner who could use AI to predict which pizzas will sell best each day, thereby reducing waste and increasing profits. This is where the concept of democratizing AI becomes crucial.
The challenge isn't that small businesses lack sufficient data for AI. Even a single pizza store generates enough data for AI to identify patterns and make useful predictions. The real issue is that small businesses can't afford to hire a dedicated AI team.
Let's delve deeper into this with an example of a T-shirt company. An accountant at the company could use AI to forecast demand, determining which memes to print on T-shirts based on what's trending on social media. A store manager could use AI to optimize product placement, enhancing the store's visual appeal and boosting sales. A buyer could use AI to decide whether to purchase a piece of fabric now or wait for a better price. A quality inspector could use AI to scan pictures of fabric for defects, ensuring only the best quality materials are used.
Each of these applications requires a unique AI project that needs to be custom-built. There's no one-size-fits-all AI solution for them. This is the challenge we face in democratizing AI: there are millions of these unique projects that no one is working on, but their combined value is colossal.
So, how do we make AI accessible to everyone? The answer lies in new AI development platforms that focus on data rather than code. Instead of writing pages of code, users provide data to train the AI. For instance, a fabric inspector could upload pictures of fabric and show the AI what defects look like. The AI learns from this data, and the inspector can adjust the data to make the AI smarter.
This approach makes AI accessible to anyone who can provide data, not just those who can code. It empowers everyone - accountants, store managers, buyers, quality inspectors, and even pizza store owners - to build their own AI systems.
Just as widespread literacy revolutionized society, democratizing AI will have a profound impact. It will create wealth and productivity, and by making it accessible to everyone, we can ensure that this wealth is spread across society.
In the coming era of AI, we won't rely on a few "high priests" to build AI systems for us. Instead, we'll empower everyone to build AI systems for themselves. This is an incredibly exciting