The brief history of artificial intelligence: The world has changed fast – what might be next? Just 10 years ago, no machine could reliably provide language or image recognition at a human level. But, as the chart shows, AI systems have become steadily more capable and are now beating humans in tests in all these domains. This rapid advancement in AI capabilities can be attributed to three fundamental factors: training computation, algorithms, and input data used for training.

The brief history of artificial intelligence: The world has changed fast – what might be next? Just 10 years ago, no machine could reliably provide language or image recognition at a human level. But, as the chart shows, AI systems have become steadily more capable and are now beating humans in tests in all these domains. This rapid advancement in AI capabilities can be attributed to three fundamental factors: training computation, algorithms, and input data used for training.

Training computation, measured in floating point operations (FLOP), is crucial for AI systems that rely on machine learning. One FLOP is equivalent to one addition, subtraction, multiplication, or division of two decimal numbers. For the first six decades, training computation increased in line with Moore's Law, doubling roughly every 20 months. However, since 2010, this exponential growth has accelerated, with a doubling time of just about 6 months. This exponential increase in training computation has played a significant role in driving the capabilities of AI systems.

Alongside training computation, the algorithms used in AI systems have also improved over time. AI researchers have developed more sophisticated and efficient algorithms that have enhanced the performance of AI systems. These algorithms are designed to process and analyze vast amounts of data, enabling AI systems to make accurate predictions and decisions.

The third factor contributing to the advancement of AI is the input data used for training. AI systems rely on large datasets to learn and improve their performance. The availability of diverse and high-quality datasets has significantly contributed to the development of AI systems that can surpass human-level performance in various tasks.

Looking towards the future, many AI experts believe that there is a real chance that human-level artificial intelligence will be developed within the next few decades. Some even believe that it could be achieved much sooner. The rapid progress in AI capabilities, coupled with ongoing research and investment, suggests that transformative AI may become a reality by 2040.

In order to harness the potential of AI, it is crucial to focus on building AI-first products. The goal is to create products that are designed with AI models from day one, by entrepreneurs who understand both the capabilities of these models and the needs and preferences of users. While research is important, creating a product that people love on a new platform requires creativity and ingenuity. It is not just about raising money and scaling up training clusters, but about creating engaging and useful AI-driven products.

One initiative that supports the development of AI-first products is AI Grant. AI Grant aims to find and support entrepreneurs who are exploring the frontier of AI. They recognize the importance of technical and pragmatic founders who have a passion for building great products. The program focuses on AI products and investments, rather than just papers and grants, as they believe that UX innovation in the AI space is just getting started.