Modern economy was built on the principle of scaling: converting variable cost into fixed cost and thus creating operating leverage. Instead of hand-writing books, we set up a printing press, decimating the variable cost of producing a copy. We have become incredibly efficient at scaling any type of manufacturing.
However, we have not yet cracked the code on scaling knowledge work. Even if its output can be scaled at almost-zero variable cost thanks to the Internet, its process - the actual process of transforming data into information - remains in the pre-industrial age. Every time a new employee joins the company, they need to be singularly trained for their tasks. Should they become unavailable for work, their skills cannot be used anymore. Training a replacement takes just as much time.
Artificial intelligence holds the promise of changing that by digitalizing cognitive operations. To get a feeling for its potential, imagine how much time it takes to train a chess grandmaster: years, if not decades. A generative adversarial network can reach that level in a matter of hours (Guardian).
The key to such impressive results lies in large training sets. The more data a neural network is able to process, the more useful it becomes. This is a potentially invaluable source of scaling potential. See, for example, how OpenAI's neural network does creative writing after processing a vast amount of text on the Internet (Talk to a Transformer).
Application-specific cognitive robots are now becoming widely available to enterprise customers For example, Automation Anywhere's IQBot features pre-trained AI models for document processing. Trained on millions of data points, it essentially centralizes the cost of learning, thus creating operating leverage.
Imagine a world where each Professional could be directly connected to a central knowledge repository. Every time one Professional learned something new, their knowledge could instantly be shared by every other Professional in their industry. AI makes this possible, potentially in the near future.
If certain knowledge work tasks (e.g. analyzing contracts) are performed by cognitive robots connected to a central learning instance, each enterprise can customize their robot to their needs and treat it like a singular digital worker (incl. data privacy and security), while the learning is shared within the network. As the robot is continuously trained, its shared core model becomes more and more useful for every participant in the ecosystem, creating exponential productivity gains.
Ultimately, F-ONE is in the business of increasing human productivity. And we are excited about the ways it can be achieved.