Ready to learn Machine Learning? Browse courses like Uncertain Knowledge and Reasoning in Artificial Intelligence developed by industry thought leaders and Experfy in Harvard Innovation Lab.
Each of us is defined and shaped by a combination of our experience and our genetic makeup. The precise balance of nature and nurture is not clear, but it is probably fair to say that as we get older, experience may become more important. This is crucial to being able to adjust and innovate because it affects our ability to adapt to circumstances.
We can obtain information to inform our decisions from a wide range of sources: direct experience, trial and error, study, training, and reading or reviewing material. Few of us would argue, however, that the best decisions are made when we have a wide range of sources of information, and can compare and contrast, drawing knowledge from all of them to inform the decision. As universities are constantly reminding their students, it is important to consider information that does NOT support your views, as well as any that does. You have to understand both sides of the argument, and not just one, to make an informed decision.
Therefore, it should come as no surprise that analytical software also needs a wide range of good quality, reliable data. It should be as detailed, wide and diverse as possible. This is probably even truer when we consider artificial intelligence (AI) systems because these are modelled on humans.
A question of time
All data are not, however, created equal. Or rather, they may be created equal, but their value changes over time, and crucially, it decays. This means, roughly speaking, that in this new industrial revolution, data are best used when fresh. This has become increasingly important over the last few years, as the volume of data has increased, and the ability to analyse it earlier has become ubiquitous.
Companies used to run analytics processes every few weeks or months, as new data became available. Now, with data from digital sources and IoT-connected devices, new data are arriving continuously, in an endless stream. There is still a place for time-series data – particularly, for example, in looking at customer behaviour – but the world of analytics is changing.
New developments in infrastructure, technology and connected devices mean we are able to start to analyse data at a much earlier point, giving us valuable insights sooner. You could almost argue that the sheer volume and velocity of the information, and proliferation of data sources, means that we have no choice but to do so, or we will be overwhelmed by data. This, in essence, is the reason we created edge analytics.
This is hugely exciting. For perhaps the first time, analytics is cool . Every day, there are news stories about exciting applications for analytics in robotics or on the edge of IoT. In fact, whenever you hear the term “ artificial intelligence ” or AI, it really means advanced analytics. Analytics is helping to improve the quality of life for people with diseases, and for countless people in less developed parts of the world. It is reducing costs or helping control consumption of energy or managing devices in the home, as well as reducing industrial costs, improving energy efficiency, and making goods cheaper as a result.
Whenever you hear the term artificial intelligence, it really means advanced analytics. For perhaps the first time, analytics is cool.
This is partly the consequence of the amount of data available. But it is only really possible because analytics, too, has changed. It is no longer the privileged reserve of data scientists, sitting in computer rooms crunching numbers on request. Instead, with new advancements, pretty much anyone with a reasonable understanding can do their own analytics.
There are single, unified interfaces that combine all the steps of the analytics life cycle, making it much easier to access and use analytics. Cloud computing means that solutions and platforms scale to meet needs, which reduces the cost of the initial implementation . These solutions also have the technical capacity required to analyse millions of data points in stream per second, and this can be done at the edge, in the cloud, in a traditional data centre or within hybrid combinations of all three.
Evolution or revolution?
We talk about digital transformation, but I prefer to think of the process as akin to evolution. The availability of data and tools to manage analytics enables those with the right capacity and ability to take advantage: a process of “technical selection,” perhaps. Looking to the future, the IoT systems and artificial intelligence platforms that we create today will become the baseline for the way future generations think about and engage with technology. They will form the foundations for future ecosystems.