Today, humans and even more so, machines, are creating more data in one week than the entire amount of data that was created by humanity since the beginning of history. And, the week after that, the amount of data will again double the entire amount of human knowledge. Combined with the ability to transmit this data at the speed of light in a fiber cable, the scope and speed of data available for decision making has gone beyond human capacity and entered nanospace.
While we are amassing all this data, we worry about how secure it is. Despite our concerns (and even the coming nanocrises), the advantages of big data are far more promising than the attendant perils. Incoming data that would seem like a tsunami to even the most adroit human statistician strikes a thirsty AI as a mere droplet. When an intelligent system drinks in those oceans of data, it makes wonderful discoveries that can benefit us all.
Data arrives in the system from three sources: human input, sensor readings, and the results of other AI operations.
Data is stored in data lakes—aggregates of data on local servers, remote servers (including the cloud), within devices, or on removable media such as flash drives and CDs.
Each data point is accessed for both quailty and utility. Any data point may be erroneous or irrelevant.
An algorithm is a process designed to compute a result in service of a defined objective. While each algorithm has many steps, it will not necessarily take all these steps to solve a problem. Because algorithms are written by humans, any step may contain unintended bias. Anticipant leaders work to ensure that all algorithms used by their organizations produce fair results.
Each result in turn is stored in a location accessible to the next entity in the chain, including other computing devices, robots, and kinetic machines.
Armed with the result, the succeeding entity takes a desired action, whether a physical movement or a further computation.
In every discipline, great leaps of understanding are being made daily by AIs as they sift through the complexity before them and identify repeating patterns. Once a pattern is associated routinely with an outcome, AI can predict that very outcome every time a known pattern appears. This so-called predictive analysis is beyond human capacity (taking place as it does in the suprahuman domain of machines) and is often beyond human comprehension. While the content for these communications is created by humans, the data imbedded in all of them is collected and shared by machines alone.